Files
obikmer/doc/theory/kmer_indexing/kmers/index.html
T
Eric Coissac 33d7a59dd9 docs: restructure documentation for kmer indexing and phylogeny theory
Reorganize the documentation structure to better separate kmer indexing logic from phylogenetic analysis methods. This involves moving several existing pages into new directories under `theory/kmer_indexing` and `theory/phylogeny`, updating navigation links across the site, and adding new content for VCF pseudo-alignment and specific phylogenetic models.

- Move `doc/index.html` to `doc/theory/kmer_indexing/minimizer_selection/index.html` to document minimizer selection logic.
- Move `doc/installation/index.html` to `doc/theory/phylogeny/kmer_based/kmer_set_distances/index.html` for whole k-mer set distance documentation.
- Move `doc/usage/dump/index.html` to `doc/theory/kmer_indexing/superkmers/index.html` to cover canonical super-kmers.
- Move `doc/usage/query/index.html` to `doc/theory/kmer_indexing/entropy_filter/index.html` for low-complexity kmer filtering details.
- Move `doc/theory/encoding/index.html` to `doc/theory/kmer_indexing/encoding/index.html` for DNA and kmer encoding specifics.
- Move `doc/theory/indexing_architecture/index.html` to `doc/theory/kmer_indexing/indexing_architecture/index.html` for partitioning and routing architecture.
- Move `doc/usage/query/index.html` (duplicate move noted in facts, likely referring to the query command usage or similar) to `doc/theory/phylogeny/snp_based/sankoff_model/index.html` for the 16-state Sankoff model.
- Move `doc/theory/minimizer_selection/index.html` to `doc/theory/phylogeny/snp_based/central_snp_model/index.html` for central-position SNP models and substitution corrections.
- Move `doc/theory/indexing_architecture/index.html` (duplicate move noted) to `doc/theory/phylogeny/snp_based/gamma_rate_heterogeneity/index.html` for gamma rate heterogeneity and automatic alpha estimation.
- Move `doc/theory/entropy_filter/index.html` to `doc/theory/phylogeny/snp_based/sampling_design/index.html` for entropy-biased sampling design.
- Move `doc/formats/index_layout/index.html` to `doc/formats/vcf/index.html` to document VCF pseudo-alignment output and structure.
- Add `doc/molecular-ecology-modified.csl` with a new XML style definition for Molecular Ecology Modified citations.
- Update `UserDocMD/formats/vcf.md` to clarify that `--vcf` implies `--pseudo-alignment`, requires `--n-sites`, and generates files alongside `--fasta`.
- Add three new BibTeX entries in `UserDocMD/references.bib` for Karlin (1997, 1995) and Bennett & Moran (2013).
- Update `UserDocMD/theory/kmer_indexing/kmers.md` to use binomial distribution for probability calculations and define k-mer sharing deviation.
- Clarify in `UserDocMD/theory/phylogeny/kmer_based/kmer_set_distances.md` that counts are used instead of presence/absence for $L^1$ distance.
- Update `UserDocMD/theory/phylogeny/snp_based/gamma_rate_heterogeneity.md` to attribute method-of-moments $\alpha$ estimation to obikmer, not Jin & Nei (1990).
- Update `UserDocMD/theory/phylogeny/snp_based/sampling_design.md` to redirect entropy sampling references to the correct phylo section.
- Add new options `--cpu-max`, `--output`, `--pair-balance`, `--sibling-stats`, `--family-overlap`, and `--shannon` to `UserDocMD/usage/phylo.md`.
- Mark SNP-based distance saving and session reuse as unavailable in `src/obikmer/src/cmd/phylo/args/distance.rs`.
2026-11-19 17:47:59 +01:00

2250 lines
78 KiB
HTML

<!doctype html>
<html lang="en" class="no-js">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width,initial-scale=1">
<link rel="prev" href="../encoding/">
<link rel="next" href="../minimizer_selection/">
<link rel="icon" href="../../../assets/images/favicon.png">
<meta name="generator" content="mkdocs-1.6.1, mkdocs-material-9.7.6">
<title>Kmers - obikmer — User Guide</title>
<link rel="stylesheet" href="../../../assets/stylesheets/main.484c7ddc.min.css">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Roboto:300,300i,400,400i,700,700i%7CRoboto+Mono:400,400i,700,700i&display=fallback">
<style>:root{--md-text-font:"Roboto";--md-code-font:"Roboto Mono"}</style>
<script>__md_scope=new URL("../../..",location),__md_hash=e=>[...e].reduce(((e,_)=>(e<<5)-e+_.charCodeAt(0)),0),__md_get=(e,_=localStorage,t=__md_scope)=>JSON.parse(_.getItem(t.pathname+"."+e)),__md_set=(e,_,t=localStorage,a=__md_scope)=>{try{t.setItem(a.pathname+"."+e,JSON.stringify(_))}catch(e){}}</script>
</head>
<body dir="ltr">
<input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
<input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
<label class="md-overlay" for="__drawer"></label>
<div data-md-component="skip">
<a href="#k-mer-size" class="md-skip">
Skip to content
</a>
</div>
<div data-md-component="announce">
</div>
<header class="md-header md-header--shadow" data-md-component="header">
<nav class="md-header__inner md-grid" aria-label="Header">
<a href="../../.." title="obikmer — User Guide" class="md-header__button md-logo" aria-label="obikmer — User Guide" data-md-component="logo">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a3 3 0 0 0 3-3 3 3 0 0 0-3-3 3 3 0 0 0-3 3 3 3 0 0 0 3 3m0 3.54C9.64 9.35 6.5 8 3 8v11c3.5 0 6.64 1.35 9 3.54 2.36-2.19 5.5-3.54 9-3.54V8c-3.5 0-6.64 1.35-9 3.54"/></svg>
</a>
<label class="md-header__button md-icon" for="__drawer">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M3 6h18v2H3zm0 5h18v2H3zm0 5h18v2H3z"/></svg>
</label>
<div class="md-header__title" data-md-component="header-title">
<div class="md-header__ellipsis">
<div class="md-header__topic">
<span class="md-ellipsis">
obikmer — User Guide
</span>
</div>
<div class="md-header__topic" data-md-component="header-topic">
<span class="md-ellipsis">
Kmers
</span>
</div>
</div>
</div>
<script>var palette=__md_get("__palette");if(palette&&palette.color){if("(prefers-color-scheme)"===palette.color.media){var media=matchMedia("(prefers-color-scheme: light)"),input=document.querySelector(media.matches?"[data-md-color-media='(prefers-color-scheme: light)']":"[data-md-color-media='(prefers-color-scheme: dark)']");palette.color.media=input.getAttribute("data-md-color-media"),palette.color.scheme=input.getAttribute("data-md-color-scheme"),palette.color.primary=input.getAttribute("data-md-color-primary"),palette.color.accent=input.getAttribute("data-md-color-accent")}for(var[key,value]of Object.entries(palette.color))document.body.setAttribute("data-md-color-"+key,value)}</script>
</nav>
</header>
<div class="md-container" data-md-component="container">
<main class="md-main" data-md-component="main">
<div class="md-main__inner md-grid">
<div class="md-sidebar md-sidebar--primary" data-md-component="sidebar" data-md-type="navigation" >
<div class="md-sidebar__scrollwrap">
<div class="md-sidebar__inner">
<nav class="md-nav md-nav--primary" aria-label="Navigation" data-md-level="0">
<label class="md-nav__title" for="__drawer">
<a href="../../.." title="obikmer — User Guide" class="md-nav__button md-logo" aria-label="obikmer — User Guide" data-md-component="logo">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a3 3 0 0 0 3-3 3 3 0 0 0-3-3 3 3 0 0 0-3 3 3 3 0 0 0 3 3m0 3.54C9.64 9.35 6.5 8 3 8v11c3.5 0 6.64 1.35 9 3.54 2.36-2.19 5.5-3.54 9-3.54V8c-3.5 0-6.64 1.35-9 3.54"/></svg>
</a>
obikmer — User Guide
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../.." class="md-nav__link">
<span class="md-ellipsis">
Home
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../installation/" class="md-nav__link">
<span class="md-ellipsis">
Installation
</span>
</a>
</li>
<li class="md-nav__item md-nav__item--active md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_3" checked>
<label class="md-nav__link" for="__nav_3" id="__nav_3_label" tabindex="0">
<span class="md-ellipsis">
Theory
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="1" aria-labelledby="__nav_3_label" aria-expanded="true">
<label class="md-nav__title" for="__nav_3">
<span class="md-nav__icon md-icon"></span>
Theory
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item md-nav__item--active md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_3_1" checked>
<label class="md-nav__link" for="__nav_3_1" id="__nav_3_1_label" tabindex="0">
<span class="md-ellipsis">
Kmer indexing
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="2" aria-labelledby="__nav_3_1_label" aria-expanded="true">
<label class="md-nav__title" for="__nav_3_1">
<span class="md-nav__icon md-icon"></span>
Kmer indexing
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../encoding/" class="md-nav__link">
<span class="md-ellipsis">
DNA encoding
</span>
</a>
</li>
<li class="md-nav__item md-nav__item--active">
<input class="md-nav__toggle md-toggle" type="checkbox" id="__toc">
<label class="md-nav__link md-nav__link--active" for="__toc">
<span class="md-ellipsis">
Kmers
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<a href="./" class="md-nav__link md-nav__link--active">
<span class="md-ellipsis">
Kmers
</span>
</a>
<nav class="md-nav md-nav--secondary" aria-label="Table of contents">
<label class="md-nav__title" for="__toc">
<span class="md-nav__icon md-icon"></span>
Table of contents
</label>
<ul class="md-nav__list" data-md-component="toc" data-md-scrollfix>
<li class="md-nav__item">
<a href="#representation-limit" class="md-nav__link">
<span class="md-ellipsis">
Representation limit
</span>
</a>
<nav class="md-nav" aria-label="Representation limit">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#k-mer-specificity" class="md-nav__link">
<span class="md-ellipsis">
K-mer specificity
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#estimating-false-discovery-rate-on-homology-inference" class="md-nav__link">
<span class="md-ellipsis">
Estimating False discovery rate on homology inference
</span>
</a>
<nav class="md-nav" aria-label="Estimating False discovery rate on homology inference">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#the-random-model" class="md-nav__link">
<span class="md-ellipsis">
The random model
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#exact-tail-probability" class="md-nav__link">
<span class="md-ellipsis">
Exact tail probability
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#odd-k-mer-lengths" class="md-nav__link">
<span class="md-ellipsis">
Odd k-mer lengths
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#karlin-ost-1988-formula" class="md-nav__link">
<span class="md-ellipsis">
Karlin--Ost (1988) formula
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#chance-sharing-across-genomes-and-the-fdr-of-homology" class="md-nav__link">
<span class="md-ellipsis">
Chance sharing across genomes and the FDR of homology
</span>
</a>
<nav class="md-nav" aria-label="Chance sharing across genomes and the FDR of homology">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#the-null-model" class="md-nav__link">
<span class="md-ellipsis">
The null model
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#exact-tail-probability_1" class="md-nav__link">
<span class="md-ellipsis">
Exact tail probability
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#approximation-with-equal-genome-sizes" class="md-nav__link">
<span class="md-ellipsis">
Approximation with equal genome sizes
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#false-discovery-rate" class="md-nav__link">
<span class="md-ellipsis">
False discovery rate
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#limits-of-the-model" class="md-nav__link">
<span class="md-ellipsis">
Limits of the model
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#1-fdr-dun-ensemble-de-k-mers-selectionnes" class="md-nav__link">
<span class="md-ellipsis">
1. FDR d'un ensemble de k-mers sélectionnés
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#2-markov-dordre-0-avec-frequences-inegales" class="md-nav__link">
<span class="md-ellipsis">
2. Markov d'ordre 0 avec fréquences inégales
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="../minimizer_selection/" class="md-nav__link">
<span class="md-ellipsis">
Minimizer selection
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../superkmers/" class="md-nav__link">
<span class="md-ellipsis">
Super-kmers
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../indexing_architecture/" class="md-nav__link">
<span class="md-ellipsis">
Partitioning and indexing architecture
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../entropy_filter/" class="md-nav__link">
<span class="md-ellipsis">
Low-complexity kmer filter
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_3_2" >
<label class="md-nav__link" for="__nav_3_2" id="__nav_3_2_label" tabindex="0">
<span class="md-ellipsis">
Phylogeny
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="2" aria-labelledby="__nav_3_2_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_3_2">
<span class="md-nav__icon md-icon"></span>
Phylogeny
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_3_2_1" >
<label class="md-nav__link" for="__nav_3_2_1" id="__nav_3_2_1_label" tabindex="0">
<span class="md-ellipsis">
Kmer-based
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="3" aria-labelledby="__nav_3_2_1_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_3_2_1">
<span class="md-nav__icon md-icon"></span>
Kmer-based
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../phylogeny/kmer_based/kmer_set_distances/" class="md-nav__link">
<span class="md-ellipsis">
Whole k-mer set distances
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_3_2_2" >
<label class="md-nav__link" for="__nav_3_2_2" id="__nav_3_2_2_label" tabindex="0">
<span class="md-ellipsis">
SNP-based
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="3" aria-labelledby="__nav_3_2_2_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_3_2_2">
<span class="md-nav__icon md-icon"></span>
SNP-based
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../phylogeny/snp_based/central_snp_model/" class="md-nav__link">
<span class="md-ellipsis">
Central-position SNP model
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../phylogeny/snp_based/gamma_rate_heterogeneity/" class="md-nav__link">
<span class="md-ellipsis">
Gamma rate heterogeneity
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../phylogeny/snp_based/sankoff_model/" class="md-nav__link">
<span class="md-ellipsis">
Sankoff model
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../phylogeny/snp_based/sampling_design/" class="md-nav__link">
<span class="md-ellipsis">
Sampling design
</span>
</a>
</li>
</ul>
</nav>
</li>
</ul>
</nav>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_4" >
<label class="md-nav__link" for="__nav_4" id="__nav_4_label" tabindex="0">
<span class="md-ellipsis">
Usage
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="1" aria-labelledby="__nav_4_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_4">
<span class="md-nav__icon md-icon"></span>
Usage
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../../usage/superkmer/" class="md-nav__link">
<span class="md-ellipsis">
superkmer
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/index_command/" class="md-nav__link">
<span class="md-ellipsis">
index
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/merge/" class="md-nav__link">
<span class="md-ellipsis">
merge
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/filter/" class="md-nav__link">
<span class="md-ellipsis">
filter
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/select/" class="md-nav__link">
<span class="md-ellipsis">
select
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/query/" class="md-nav__link">
<span class="md-ellipsis">
query
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/dump/" class="md-nav__link">
<span class="md-ellipsis">
dump
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/annotate/" class="md-nav__link">
<span class="md-ellipsis">
annotate
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/phylo/" class="md-nav__link">
<span class="md-ellipsis">
phylo
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/unitig/" class="md-nav__link">
<span class="md-ellipsis">
unitig
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/estimate/" class="md-nav__link">
<span class="md-ellipsis">
estimate
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/convert/" class="md-nav__link">
<span class="md-ellipsis">
convert
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/utils/" class="md-nav__link">
<span class="md-ellipsis">
utils
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/pack/" class="md-nav__link">
<span class="md-ellipsis">
pack
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../usage/predicates/" class="md-nav__link">
<span class="md-ellipsis">
Predicates and taxonomy paths
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item md-nav__item--nested">
<input class="md-nav__toggle md-toggle " type="checkbox" id="__nav_5" >
<label class="md-nav__link" for="__nav_5" id="__nav_5_label" tabindex="0">
<span class="md-ellipsis">
Formats
</span>
<span class="md-nav__icon md-icon"></span>
</label>
<nav class="md-nav" data-md-level="1" aria-labelledby="__nav_5_label" aria-expanded="false">
<label class="md-nav__title" for="__nav_5">
<span class="md-nav__icon md-icon"></span>
Formats
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="../../../formats/index_layout/" class="md-nav__link">
<span class="md-ellipsis">
Index construction and on-disk layout
</span>
</a>
</li>
<li class="md-nav__item">
<a href="../../../formats/vcf/" class="md-nav__link">
<span class="md-ellipsis">
VCF pseudo-alignment
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="../../../architecture/" class="md-nav__link">
<span class="md-ellipsis">
Architecture notes
</span>
</a>
</li>
</ul>
</nav>
</div>
</div>
</div>
<div class="md-sidebar md-sidebar--secondary" data-md-component="sidebar" data-md-type="toc" >
<div class="md-sidebar__scrollwrap">
<div class="md-sidebar__inner">
<nav class="md-nav md-nav--secondary" aria-label="Table of contents">
<label class="md-nav__title" for="__toc">
<span class="md-nav__icon md-icon"></span>
Table of contents
</label>
<ul class="md-nav__list" data-md-component="toc" data-md-scrollfix>
<li class="md-nav__item">
<a href="#representation-limit" class="md-nav__link">
<span class="md-ellipsis">
Representation limit
</span>
</a>
<nav class="md-nav" aria-label="Representation limit">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#k-mer-specificity" class="md-nav__link">
<span class="md-ellipsis">
K-mer specificity
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#estimating-false-discovery-rate-on-homology-inference" class="md-nav__link">
<span class="md-ellipsis">
Estimating False discovery rate on homology inference
</span>
</a>
<nav class="md-nav" aria-label="Estimating False discovery rate on homology inference">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#the-random-model" class="md-nav__link">
<span class="md-ellipsis">
The random model
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#exact-tail-probability" class="md-nav__link">
<span class="md-ellipsis">
Exact tail probability
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#odd-k-mer-lengths" class="md-nav__link">
<span class="md-ellipsis">
Odd k-mer lengths
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#karlin-ost-1988-formula" class="md-nav__link">
<span class="md-ellipsis">
Karlin--Ost (1988) formula
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#chance-sharing-across-genomes-and-the-fdr-of-homology" class="md-nav__link">
<span class="md-ellipsis">
Chance sharing across genomes and the FDR of homology
</span>
</a>
<nav class="md-nav" aria-label="Chance sharing across genomes and the FDR of homology">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#the-null-model" class="md-nav__link">
<span class="md-ellipsis">
The null model
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#exact-tail-probability_1" class="md-nav__link">
<span class="md-ellipsis">
Exact tail probability
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#approximation-with-equal-genome-sizes" class="md-nav__link">
<span class="md-ellipsis">
Approximation with equal genome sizes
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#false-discovery-rate" class="md-nav__link">
<span class="md-ellipsis">
False discovery rate
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#limits-of-the-model" class="md-nav__link">
<span class="md-ellipsis">
Limits of the model
</span>
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#1-fdr-dun-ensemble-de-k-mers-selectionnes" class="md-nav__link">
<span class="md-ellipsis">
1. FDR d'un ensemble de k-mers sélectionnés
</span>
</a>
</li>
<li class="md-nav__item">
<a href="#2-markov-dordre-0-avec-frequences-inegales" class="md-nav__link">
<span class="md-ellipsis">
2. Markov d'ordre 0 avec fréquences inégales
</span>
</a>
</li>
</ul>
</nav>
</div>
</div>
</div>
<div class="md-content" data-md-component="content">
<article class="md-content__inner md-typeset">
<h1 id="k-mer-size">K-mer size</h1>
<p>A k-mer is a DNA sequence of length <span class="arithmatex">\(k\)</span>. OBIKmer is designed to index large k-mers. Here, <em>large</em> refers to a k-mer length for which the probability of observing the same k-mer in several genomes by chance alone is sufficiently small. The k-mer length must therefore be large enough that, when a k-mer is observed in several genomes of the dataset, the probability that this shared occurrence is due to chance alone is sufficiently low for the match to be considered evidence of homology.</p>
<p>OBIKmer restricts <span class="arithmatex">\(k\)</span> to odd values in the range</p>
<div class="arithmatex">\[
11 \leq k \leq 31.
\]</div>
<p>These limits result from two different considerations. The upper bound is imposed by the representation used by OBIKmer, whereas the lower bound defines a practical range of k-mer lengths for genome-scale sequence comparisons. This range covers most practical applications: the lower limit is well below the k-mer size that can be considered large, even for the smallest bacterial genomes, whereas the upper limit provides sufficient specificity for most applications, including those involving very large genomes.</p>
<h2 id="representation-limit">Representation limit</h2>
<p>OBIKmer uses two bits to encode each nucleotide. Therefore, a k-mer requires <span class="arithmatex">\(2k\)</span> bits. To ensure computational efficiency, OBIKmer limits the representation of k-mers to a single 64-bit word, thereby restricting their length.</p>
<div class="arithmatex">\[
2k \leq 64,
\]</div>
<p>This gives the hard representation limit: <span class="arithmatex">\(k \leq 32.\)</span> Furthermore, OBIKmer only considers odd values of <span class="arithmatex">\(k\)</span>, meaning the largest supported k-mer size is <span class="arithmatex">\(31\)</span>.</p>
<h3 id="k-mer-specificity">K-mer specificity</h3>
<p>One simple way to illustrate the effect of k-mer size is to consider a random DNA sequence in which the four nucleotides are independent and equally probable. There are <span class="arithmatex">\(4^k\)</span> possible k-mers, so a particular k-mer is expected to occur once every <span class="arithmatex">\(4^k\)</span> positions on average. In a sequence of length <span class="arithmatex">\(G\)</span>, the expected number of occurrences of a particular k-mer is approximately <span class="arithmatex">\(G \cdot 4^{-k}\)</span>. When k is large enough to ensure <span class="arithmatex">\(4^k &gt;&gt; G\)</span>, the chance of a second occurrence of the same k-mer becomes close to zero. Consequently, observing the same k-mer twice can only be explained by a homology relationship, except for low-complexity k-mers, which are overrepresented in genomes. </p>
<h3 id="estimating-false-discovery-rate-on-homology-inference">Estimating False discovery rate on homology inference</h3>
<p>OBIKmer uses k-mers long enough that their occurrence in multiple indexed genomes provides evidence of homology. The corresponding false discovery rate (FDR) can be expressed in terms of the probability of observing a k-mer in at least <span class="arithmatex">\(R\)</span> of the <span class="arithmatex">\(S\)</span> indexed genomes under a purely random model.</p>
<h4 id="the-random-model">The random model</h4>
<p>The random model, or null hypothesis, assumes that the <span class="arithmatex">\(S\)</span> genomes are unrelated and that each genome is an independent random sequence in which each nucleotide occurs with equal probability. Rejecting this hypothesis for a given k-mer indicates that its presence in multiple genomes is unlikely to be explained by chance alone, providing evidence for homology.</p>
<p>For a genome of size <span class="arithmatex">\(G\)</span>, the number of occurrences of a given canonical k-mer <span class="arithmatex">\(K\)</span> is approximated by a Poisson distribution with parameter</p>
<div class="arithmatex">\[ \lambda = 2\frac{G}{4^k}, \]</div>
<p>which is also the expected number of occurrences under the random model. The factor of two accounts for k-mer canonicalisation, which halves the k-mer space when <span class="arithmatex">\(k\)</span> is odd.</p>
<p>The probability of observing at least one occurrence of this k-mer is therefore</p>
<div class="arithmatex">\[
\begin{aligned}
P(n_K &gt; 0) &amp;= 1-e^{-\lambda} \\
&amp;= 1-e^{-2G/4^k}.
\end{aligned}
\]</div>
<p>Assuming equal genome lengths, the number of genomes containing at least one occurrence of <span class="arithmatex">\(K\)</span> among the <span class="arithmatex">\(S\)</span> sequences follows a binomial distribution. The probability of observing <span class="arithmatex">\(K\)</span> in at least <span class="arithmatex">\(r\)</span> genomes is therefore</p>
<div class="arithmatex">\[
P(X \geq r)=\sum_{j=r}^{S}\binom{S}{j}\left(1-e^{-2G/4^k}\right)^j\left(e^{-2G/4^k}\right)^{S-j}.
\]</div>
<p>This sum is the regularized incomplete beta function <span class="arithmatex">\(I_p(r,\,S-r+1)\)</span>, where</p>
<div class="arithmatex">\[
p = 1-e^{-2G/4^k}.
\]</div>
<p>Different genome sizes</p>
<p>When genome sizes differ, the probability of observing a given k-mer also differs between genomes. A simple approximation is to replace the individual genome sizes <span class="arithmatex">\(G_i\)</span> by their mean</p>
<div class="arithmatex">\[ \bar G = \frac{1}{S}\sum_{i=1}^{S}G_i. \]</div>
<p>This gives a common probability</p>
<div class="arithmatex">\[ \bar p = 1-e^{-2\bar G/4^k}, \]</div>
<p>and the number of genomes containing the k-mer can then be approximated by a binomial distribution:</p>
<div class="arithmatex">\[ P(X\geq r)=\sum_{j=r}^{S}\binom{S}{j}\bar p^j(1-\bar p)^{S-j}. \]</div>
<p>This approximation is conservative for the upper tail. Hoeffding (1956 p. 717, Eq. 26)<sup id="fnref:Hoeffding1956-bx"><a class="footnote-ref" href="#fn:Hoeffding1956-bx">1</a></sup> shows that, for a fixed mean, the binomial upper tail is at least as large as the corresponding Poisson-binomial tail for</p>
<div class="arithmatex">\[ r\geq S\bar p+1. \]</div>
<p>Using the mean genome size therefore provides a conservative estimate of the FDR.</p>
<p>When genome sizes differ substantially, however, this approximation can become overly conservative. In that case, the exact Poisson-binomial distribution can be computed directly from the individual genome-specific probabilities <span class="arithmatex">\(p_i\)</span>.</p>
<h3 id="exact-tail-probability">Exact tail probability</h3>
<p>With genome-specific probabilities <span class="arithmatex">\(p_i\)</span>, the number of genomes containing the k-mer follows a Poisson-binomial distribution. Writing the probability of exactly <span class="arithmatex">\(j\)</span> presences as a sum over all subsets of <span class="arithmatex">\(j\)</span> genomes would be impractical, since there can be up to <span class="arithmatex">\(2^S\)</span> subsets. The distribution can instead be obtained by adding the genomes one at a time.</p>
<p>Let <span class="arithmatex">\(d_j\)</span> be the probability that exactly <span class="arithmatex">\(j\)</span> of the genomes processed so far contain the k-mer. Adding genome <span class="arithmatex">\(i\)</span> leaves <span class="arithmatex">\(j\)</span> unchanged if the genome does not contain the k-mer, and moves <span class="arithmatex">\(j-1\)</span> to <span class="arithmatex">\(j\)</span> if it does:</p>
<div class="arithmatex">\[
\left \{
\begin{aligned}
d'_j &amp;= d_j(1-p_i) + d_{j-1}p_i, \\
d_0&amp;=1
\end{aligned}
\right .
\]</div>
<p>before processing the first genome.</p>
<p>The table <span class="arithmatex">\(d\)</span> only needs to record how many genomes contain the k-mer, not which ones, reducing the computational cost to <span class="arithmatex">\(O(S^2)\)</span>. The result is independent of the order in which the genomes are processed and of the k-mer itself, so it can be computed once per index. The upper tail is then</p>
<div class="arithmatex">\[
P(X\geq r)=1-\sum_{j=0}^{r-1}d_j.
\]</div>
<p>Only <span class="arithmatex">\(d_0,\ldots,d_{r-1}\)</span> are needed, reducing the cost to <span class="arithmatex">\(O(Sr)\)</span>.</p>
<p>When all <span class="arithmatex">\(p_i\)</span> are small, as is typically the case for large <span class="arithmatex">\(k\)</span>, the Poisson-binomial distribution can also be approximated by a Poisson distribution with mean</p>
<div class="arithmatex">\[
\lambda=\sum_{i=1}^{S}p_i.
\]</div>
<p>The approximation error is bounded in terms of <span class="arithmatex">\(\sum_i p_i^2\)</span> by Le Cam's inequality (<strong>LeCam1960-pe?</strong>).</p>
<hr />
<p>This probability decreases by a factor of four for every additional nucleotide in the k-mer. Conversely, it increases linearly with the amount of sequence searched.</p>
<p>This simple model already shows why very short k-mers provide limited specificity for genome-scale analyses. With <span class="arithmatex">\(k=11\)</span>,</p>
<div class="arithmatex">\[
4^{11}\simeq4.2\times10^6,
\]</div>
<p>which is a small combinatorial space even compared with the size of very small bacterial genomes. In contrast,</p>
<div class="arithmatex">\[
4^{31}\simeq4.6\times10^{18},
\]</div>
<p>which is vastly larger than the number of k-mer positions even in very large genomes. Thus, increasing <span class="arithmatex">\(k\)</span> from 11 to 31 provides an enormous increase in sequence specificity.</p>
<p>The estimate above is intentionally simplistic. Real genomes do not consist of independent, equiprobable nucleotides: nucleotide composition and sequence dependencies affect the probability of observing particular words. More accurate estimates of the probability of observing matches by chance therefore take sequence length and nucleotide composition into account.</p>
<p>Karlin and Ost provide an analytical framework for this problem based on the distribution of the longest exact match expected between random sequences. If <span class="arithmatex">\(L_{\max}\)</span> denotes the length of the longest seed observed in random sequences of given size and nucleotide composition, the corresponding seed-level probability is</p>
<div class="arithmatex">\[
P_{\mathrm{seeds}}(L)=P(L_{\max}\geq L).
\]</div>
<p>Conversely, for a chosen false-positive probability <span class="arithmatex">\(P\)</span>, one can determine the minimum seed length <span class="arithmatex">\(L_{\min}\)</span> such that matches of length <span class="arithmatex">\(L_{\min}\)</span> or greater are not expected to occur by chance. This is the approach used by Repseek to determine statistically significant seed lengths.</p>
<p>The simple <span class="arithmatex">\(G/4^k\)</span> estimate and the Karlin–Ost statistics therefore describe the same fundamental effect at different levels of approximation: the probability of a chance match decreases rapidly with k-mer length, while the amount of sequence being searched increases the opportunity for such matches.</p>
<p>Consequently, the appropriate value of <span class="arithmatex">\(k\)</span> depends on the total amount of sequence being compared and its nucleotide composition. Choosing <span class="arithmatex">\(k\)</span> appropriately allows the false-positive rate of homology inference to be controlled.</p>
<p>Considering that bacterial genomes are typically a few megabases long, with the smallesr plublished today <em>Mycoplasma genitalium</em> having a size <span class="arithmatex">\(G=580\,Mb\)</span> (Fraser <em>et al.</em> 1995)<sup id="fnref:Fraser1995-jc"><a class="footnote-ref" href="#fn:Fraser1995-jc">2</a></sup>, and the largest <span class="arithmatex">\(13\,Mb\)</span> in <em>Sorangium cellulosum</em> (Schneiker <em>et al.</em> 2007)<sup id="fnref:Schneiker2007-gs"><a class="footnote-ref" href="#fn:Schneiker2007-gs">3</a></sup>. For such genomes, the space of 11-mers is already to small to get a reasonable false positive rate:</p>
<div class="arithmatex">\[
\begin{aligned}
P(C\geq1) &amp;= 1- e^{-\frac{580.000^2}{2 \cdot 4^{11}}} \\
&amp;\simeq 1
\end{aligned}
\]</div>
<p>, whereas 21-mers already provide a space of several trillion possible sequences.</p>
<p>Plant genomes illustrate the opposite situation. Plant genome sizes span several orders of magnitude, from the smallest angiosperm genomes present in the carnivorous plans genus (Fleischmann <em>et al.</em> 2014)<sup id="fnref:Fleischmann2014-ed"><a class="footnote-ref" href="#fn:Fleischmann2014-ed">4</a></sup> about 60 Mb to the largest one 160 Gb in Tmesipteris oblanceolata, a New Caledonian fork fern (Fernández <em>et al.</em> 2024)<sup id="fnref:Fernandez2024-kd"><a class="footnote-ref" href="#fn:Fernandez2024-kd">5</a></sup>, and large genomes often contain substantial amounts of repeated sequence. Even at <span class="arithmatex">\(k=31\)</span>, however, the theoretical k-mer space remains vastly larger than any individual genome. At these lengths, the limiting factor is therefore not exhaustion of the theoretical k-mer space but the biological repetitiveness and non-random composition of the genome.</p>
<p>For example, the number of possible k-mers grows as follows:</p>
<table>
<thead>
<tr>
<th style="text-align: right;"><span class="arithmatex">\(k\)</span></th>
<th style="text-align: right;"><span class="arithmatex">\(4^k\)</span></th>
<th style="text-align: right;"><span class="arithmatex">\(4^k/2\)</span> after canonicalization</th>
</tr>
</thead>
<tbody>
<tr>
<td style="text-align: right;">11</td>
<td style="text-align: right;"><span class="arithmatex">\(4.19\times10^6\)</span></td>
<td style="text-align: right;"><span class="arithmatex">\(2.10\times10^6\)</span></td>
</tr>
<tr>
<td style="text-align: right;">15</td>
<td style="text-align: right;"><span class="arithmatex">\(1.07\times10^9\)</span></td>
<td style="text-align: right;"><span class="arithmatex">\(5.37\times10^8\)</span></td>
</tr>
<tr>
<td style="text-align: right;">21</td>
<td style="text-align: right;"><span class="arithmatex">\(4.40\times10^{12}\)</span></td>
<td style="text-align: right;"><span class="arithmatex">\(2.20\times10^{12}\)</span></td>
</tr>
<tr>
<td style="text-align: right;">25</td>
<td style="text-align: right;"><span class="arithmatex">\(1.13\times10^{15}\)</span></td>
<td style="text-align: right;"><span class="arithmatex">\(5.63\times10^{14}\)</span></td>
</tr>
<tr>
<td style="text-align: right;">31</td>
<td style="text-align: right;"><span class="arithmatex">\(4.61\times10^{18}\)</span></td>
<td style="text-align: right;"><span class="arithmatex">\(2.31\times10^{18}\)</span></td>
</tr>
</tbody>
</table>
<p>Canonicalization approximately halves the number of possible representations because a k-mer and its reverse complement are represented by the same canonical form.</p>
<p>The distinction between the simple <span class="arithmatex">\(4^k\)</span> model and the Karlin--Ost framework is important. The former describes the size of the combinatorial k-mer space; the latter describes the probability of observing long matches in sequences whose composition may depart from the uniform random model. Consequently, the value of <span class="arithmatex">\(k\)</span> cannot be selected from genome size alone.</p>
<p>These scales have very different implications for different genome sizes. </p>
<p>The random-sequence calculation is only a null model. Real genomes exhibit strong and reproducible oligonucleotide composition biases. Karlin and colleagues showed that relative abundances of short oligonucleotides, in particular dinucleotides, form characteristic genomic signatures that remain relatively stable within genomes and differ between organisms (<strong>Karlin1995?</strong>; <strong>Karlin1997?</strong>). More generally, short k-mer frequencies can therefore capture compositional properties of genomes rather than provide highly specific sequence identifiers.</p>
<p>This distinction is important for whole-genome k-mer comparison. Increasing <span class="arithmatex">\(k\)</span> increases the theoretical diversity of the k-mer space exponentially and generally increases the proportion of k-mers that occur at unique genomic positions. Conversely, shorter k-mers occur more frequently and are more strongly influenced by genomic composition and repeated sequence. Empirical studies of k-mer spectra show this transition clearly: for example, k-mers of length <span class="arithmatex">\(k\leq11\)</span> can behave primarily as genomic-composition signatures, whereas longer k-mers progressively provide greater sequence specificity.</p>
<p>The supported range <span class="arithmatex">\(11\leq k\leq31\)</span> should therefore be understood as an engineering compromise rather than as a universal biological optimum. Smaller values increase the probability that unrelated genomic regions share the same k-mers, while larger values increase specificity but also make the representation more sensitive to substitutions, indels, and incomplete homology. The appropriate value within this range consequently depends on the biological scale and comparison task.</p>
<h2 id="odd-k-mer-lengths">Odd k-mer lengths</h2>
<p>OBIKmer additionally requires <span class="arithmatex">\(k\)</span> to be odd. This constraint is independent of the resolution argument above.</p>
<p>For an odd-length DNA sequence, a k-mer cannot be identical to its own reverse complement. Consequently, every k-mer has two distinct orientations, which can be unambiguously collapsed into a canonical representation. For even <span class="arithmatex">\(k\)</span>, self-reverse-complementary sequences exist, so a k-mer can coincide with its own reverse complement.</p>
<p>Odd <span class="arithmatex">\(k\)</span> is therefore <strong>not required for canonicalization itself</strong>: the canonical form</p>
<div class="arithmatex">\[
s^c=\min_{\mathrm{lex}}\left(s,\operatorname{RC}(s)\right)
\]</div>
<p>remains well defined for even <span class="arithmatex">\(k\)</span>. The odd-<span class="arithmatex">\(k\)</span> restriction instead guarantees that the two orientations are always distinct, which simplifies the representation and orientation handling used by OBIKmer.</p>
<h2 id="karlin-ost-1988-formula">Karlin--Ost (1988) formula</h2>
<p>(Karlin &amp; Ost 1988)<sup id="fnref:Karlin1988"><a class="footnote-ref" href="#fn:Karlin1988">6</a></sup> considers <span class="arithmatex">\(s\)</span> independent stationary letter sequences over a finite alphabet and studies <span class="arithmatex">\(K_{r,s}(N)\)</span>, the length of the longest word common to <span class="arithmatex">\(r\)</span> of the <span class="arithmatex">\(s\)</span> sequences, each of length <span class="arithmatex">\(N\)</span>. For i.i.d. equiprobable letters over an alphabet of size <span class="arithmatex">\(m\)</span> (here <span class="arithmatex">\(m=4\)</span> for DNA), the characteristic local match parameter reduces to</p>
<div class="arithmatex">\[
\lambda^{[r]} = m^{1-r},
\]</div>
<p>so for <span class="arithmatex">\(r=2\)</span> (a word shared by two occurrences), <span class="arithmatex">\(\lambda = m^{-1} = 1/4\)</span>.</p>
<p>The expected number of common words of length <span class="arithmatex">\(k\)</span> among the <span class="arithmatex">\(\binom{s}{r}\)</span> relevant groups of sequences is</p>
<div class="arithmatex">\[
S_1(N,k) = N^{r}\binom{s}{r}\gamma\,\lambda^{k}\,(1+o(1)),
\]</div>
<p>with <span class="arithmatex">\(\gamma=1\)</span> exactly in the i.i.d. equiprobable case. Theorem 2.2 gives the limiting distribution of <span class="arithmatex">\(K_{r,s}(N)\)</span>, with <span class="arithmatex">\(k\)</span> set on the scale <span class="arithmatex">\(k=\left[\dfrac{\log N^{r}}{-\log\lambda^{*}}+x+1\right]\)</span>:</p>
<div class="arithmatex">\[
\lim_{N\to\infty}\left[\Pr\{K_{r,s}(N)&lt;k\} - \exp\{-(1-\lambda^{*})\,S_1(N,k)\}\right] = 0,
\]</div>
<p>equivalently</p>
<div class="arithmatex">\[
\frac{-\log\Pr\{K_{r,s}(N)&lt;k\}}{(\lambda^{*})^{\rho(N,x)}} \xrightarrow[N\to\infty]{} (1-\lambda^{*})\,\gamma^{*}(\lambda^{*})^{x}.
\]</div>
<p>The factor <span class="arithmatex">\((1-\lambda^{*})\)</span> multiplying <span class="arithmatex">\(S_1(N,k)\)</span> is the key correction over the naive pairwise-collision count: it de-clumps overlapping candidate matches, which are not independent events, into an effective Poisson rate. This theorem is stated for <span class="arithmatex">\(s\geq2\)</span> independent sequences (e.g. comparing two distinct genomes); the intra-sequence self-repeat case (a single genome compared to itself) is treated separately, via the reverse-complement match length <span class="arithmatex">\(R(N)\)</span>, in (Karlin &amp; Ost 1985)<sup id="fnref:Karlin1985"><a class="footnote-ref" href="#fn:Karlin1985">7</a></sup>.</p>
<p><span class="arithmatex">\(o(1)\)</span> (« petit o de 1 ») désigne une quantité qui <strong>tend vers 0 quand <span class="arithmatex">\(N\to\infty\)</span></strong>, mais dont on ne précise pas la vitesse exacte ni la formule exacte. C'est une façon d'écrire « il reste un terme d'erreur, mais il devient négligeable ».</p>
<p>Concrètement dans <span class="arithmatex">\(S_1(N,k) = N^{r}\binom{s}{r}\gamma\,\lambda^{k}\,(1+o(1))\)</span> :</p>
<ul>
<li><span class="arithmatex">\(N^{r}\binom{s}{r}\gamma\,\lambda^{k}\)</span> est la valeur <strong>exacte à la limite</strong>, celle qu'on utilise en pratique (c'est elle qui redonne <span class="arithmatex">\(E[C]\)</span> de notre discussion).</li>
<li><span class="arithmatex">\((1+o(1))\)</span> dit : « la vraie valeur de <span class="arithmatex">\(S_1(N,k)\)</span> n'est pas <em>exactement</em> égale à ce produit pour <span class="arithmatex">\(N\)</span> fini, mais le rapport entre les deux tend vers 1 quand <span class="arithmatex">\(N\to\infty\)</span> ». Autrement dit <span class="arithmatex">\(S_1(N,k) / \left[N^{r}\binom{s}{r}\gamma\lambda^{k}\right] \to 1\)</span>.</li>
</ul>
<p>Pourquoi ce terme existe : la formule <span class="arithmatex">\(F_\sigma(k)/\lambda^k \to \gamma\)</span> (éq. 2.8 du papier) est elle-même une limite quand <span class="arithmatex">\(k\to\infty\)</span> — pour <span class="arithmatex">\(k\)</span> fini, il y a un petit écart entre <span class="arithmatex">\(F_\sigma(k)\)</span> et <span class="arithmatex">\(\gamma\lambda^k\)</span>. C'est cet écart-là qui devient le <span class="arithmatex">\(o(1)\)</span>.</p>
<p>Ce que ça implique pour l'usage pratique : dans le cas iid équiprobable que tu utilises (ADN, <span class="arithmatex">\(\gamma=1\)</span> exactement, pas seulement à la limite — je l'ai vérifié directement sur la dérivation : <span class="arithmatex">\(F_\sigma(k)=m^{-k}=\lambda^k\)</span> pour tout <span class="arithmatex">\(k\)</span>, pas juste asymptotiquement), le <span class="arithmatex">\(o(1)\)</span> est en réalité <strong>nul</strong> dans ce cas précis, pas juste négligeable. La formule <span class="arithmatex">\(S_1(N,k)=N^r\binom{s}{r}\lambda^k\)</span> est donc exacte (à l'asymptotique en <span class="arithmatex">\(N\)</span> près, qui elle vient d'ailleurs — de l'approximation de Poisson elle-même, pas de ce <span class="arithmatex">\(\gamma\)</span>).</p>
<p>Les deux approches sont légitimes, mais elles répondent à des questions différentes — et il y a une relation algébrique exacte entre les deux qui vaut la peine d'être explicitée.</p>
<p><strong>Ce que calcule l'approche naïve.</strong> Pour <strong>un k-mer <span class="arithmatex">\(w\)</span> déjà observé et fixé</strong>, <span class="arithmatex">\(p_i\simeq1-\exp(-G_i/4^k)\)</span> est la probabilité que <em>ce mot précis</em> apparaisse quelque part dans le génome <span class="arithmatex">\(i\)</span> par hasard, et <span class="arithmatex">\(P(R\geq r)\)</span> (Poisson-binomiale ou binomiale) est la probabilité que <em>ce mot précis</em> soit partagé par au moins <span class="arithmatex">\(r\)</span> des <span class="arithmatex">\(S\)</span> génomes par hasard. C'est une <strong>p-value par observation</strong> : « j'ai vu ce k-mer particulier dans 5 génomes sur 10, est-ce surprenant pour lui ? »</p>
<p><strong>Ce que calcule Karlin-Ost.</strong> <span class="arithmatex">\(S_1(k)\)</span> et <span class="arithmatex">\(P(K_{r,s}\geq k)\)</span> répondent à : « parmi <strong>tous les <span class="arithmatex">\(4^k\)</span> mots possibles</strong>, quelle est la probabilité qu'au moins un seul d'entre eux atteigne ce seuil de partage par hasard, n'importe où dans les génomes ? » C'est une question <strong>globale sur tout l'espace des k-mers</strong>, pas sur un mot déjà identifié.</p>
<p><strong>La relation exacte entre les deux.</strong> Reprenons le cas <span class="arithmatex">\(r=s=2\)</span>, génomes égaux de taille <span class="arithmatex">\(G\)</span> :</p>
<ul>
<li>Naïf (mot <span class="arithmatex">\(w\)</span> fixé) : <span class="arithmatex">\(P(R=2\mid w)\simeq p^2=(G\cdot4^{-k})^2=G^2\cdot4^{-2k}\)</span></li>
<li>Karlin-Ost : <span class="arithmatex">\(S_1=G^2\cdot4^{-k}\)</span></li>
</ul>
<p>Le rapport est exactement <span class="arithmatex">\(S_1 / P(R=2\mid w) = 4^{k}\)</span> — c'est-à-dire le nombre de mots possibles. Ce n'est pas une coïncidence : <span class="arithmatex">\(S_1=N^r\binom{s}{r}\lambda^k\)</span> vient de <span class="arithmatex">\(\lambda^k=\sum_w \Pr\{w\}^r\)</span>, une somme explicite sur les <span class="arithmatex">\(4^k\)</span> mots, alors que le calcul naïf tel que présenté ne somme sur rien — il traite <span class="arithmatex">\(w\)</span> comme donné.</p>
<p><strong>Donc l'approche naïve, formellement, c'est Karlin-Ost auquel il manque le facteur « nombre de k-mers testés ».</strong> Pour la rendre comparable à la question « dois-je m'attendre à <em>une</em> collision fortuite n'importe où dans mon jeu de données », il faudrait une correction à la Bonferroni : multiplier <span class="arithmatex">\(P(R\geq r\mid w)\)</span> par le nombre effectif de k-mers canoniques testables (<span class="arithmatex">\(\approx4^k/2\)</span>). Karlin-Ost fait cette correction analytiquement et correctement (avec en prime le facteur de déclumping <span class="arithmatex">\((1-\lambda)\)</span> qu'on a discuté, qui corrige le sur-comptage des matches chevauchants — une union bound brute comme le Bonferroni décrit ci-dessus le surestimerait légèrement).</p>
<p><strong>Ce qui veut dire, en pratique, que les deux outils ne sont pas concurrents mais complémentaires :</strong>
- <strong>Choix a priori de <span class="arithmatex">\(k\)</span> / FPR global du dataset</strong> (la question de <code>kmers.md</code> depuis le début de cette conversation) → Karlin-Ost, parce qu'il intègre nativement la multiplicité sur tous les mots possibles.
- <strong>Score de significativité a posteriori pour un k-mer effectivement observé dans l'index</strong> (« ce hit précis, présent dans <span class="arithmatex">\(r\)</span> génomes sur <span class="arithmatex">\(S\)</span>, est-il compatible avec le hasard ? ») → l'approche Poisson-binomiale est exactement le bon outil, et même préférable à Karlin-Ost pour cet usage : elle gère nativement des <span class="arithmatex">\(G_i\)</span> inégaux sans les subtilités de convergence du théorème 2.2, et elle donne directement une p-value par hit exploitable pour filtrer les résultats d'un index présence/absence — ce qui correspond bien à l'usage qu'a en tête ton correspondant.</p>
<p>Une seule réserve sur le texte proposé tel quel : <span class="arithmatex">\(p_i\simeq1-\exp(-G_i/4^k)\)</span> souffre du même biais que celui identifié plus tôt dans notre discussion — un mot de longueur <span class="arithmatex">\(k\)</span> n'est pas indépendant à chaque position (chevauchement), donc ce <span class="arithmatex">\(p_i\)</span> est un peu optimiste (légèrement trop haut) pour les mêmes raisons que <span class="arithmatex">\(E[C]\)</span> l'était. Négligeable pour <span class="arithmatex">\(k\geq21\)</span>, plus discutable à <span class="arithmatex">\(k=11\)</span>.</p>
<h2 id="chance-sharing-across-genomes-and-the-fdr-of-homology">Chance sharing across genomes and the FDR of homology</h2>
<p>The formulas above ask how likely a match is between two sequences. The question OBIKmer actually has to answer is different: a k-mer is found in <span class="arithmatex">\(r\)</span> of the <span class="arithmatex">\(S\)</span> indexed genomes, and one wants to know whether this sharing can be explained by chance or whether it is evidence of homology. This section builds a null model for that question and turns it into a false discovery rate (FDR) for the call "a k-mer shared by at least <span class="arithmatex">\(r\)</span> genomes is homologous".</p>
<h3 id="the-null-model">The null model</h3>
<p>The null hypothesis is that the <span class="arithmatex">\(S\)</span> genomes are unrelated, that is, that each one is an independent random sequence. Rejecting it for a k-mer means that chance alone cannot explain why the k-mer is shared, which is the signal of homology one is looking for.</p>
<p>The model only records whether a k-mer is present in a genome, not how many times. Two consequences follow. First, repeats inside a genome are irrelevant, and the size <span class="arithmatex">\(G_i\)</span> of genome <span class="arithmatex">\(i\)</span> is its number of <strong>distinct canonical k-mers</strong>. Second, a repeat family shared between genomes is paralogy, which is a form of homology; it is a true positive and not a false one. The only k-mers that could be shared by composition rather than by descent are low-complexity ones, which OBIKmer removes beforehand with its entropy filter.</p>
<p>Everything takes place in canonical space. Because <span class="arithmatex">\(k\)</span> is odd, no k-mer equals its own reverse complement, so the <span class="arithmatex">\(4^k\)</span> k-mers pair up perfectly and the space contains exactly</p>
<div class="arithmatex">\[
N=\frac{4^k}{2}
\]</div>
<p>canonical k-mers. A given canonical k-mer is present in genome <span class="arithmatex">\(i\)</span> by chance with probability</p>
<div class="arithmatex">\[
p_i = 1-\left(1-\frac1N\right)^{G_i}\simeq \frac{G_i}{N}=\frac{2\,G_i}{4^k}.
\]</div>
<p>This is twice the <span class="arithmatex">\(G/4^k\)</span> used earlier in this page, because there <span class="arithmatex">\(G\)</span> is a number of positions on one strand, whereas <span class="arithmatex">\(G_i\)</span> here counts canonical k-mers, for which both strands are already merged.</p>
<p>Let <span class="arithmatex">\(R\)</span> be the number of genomes, out of <span class="arithmatex">\(S\)</span>, that contain a given canonical k-mer by chance. Genomes are independent under the null, so <span class="arithmatex">\(R\)</span> is a sum of independent presence indicators, each with its own probability <span class="arithmatex">\(p_i\)</span>.</p>
<h3 id="exact-tail-probability_1">Exact tail probability</h3>
<p>When genome sizes differ, the <span class="arithmatex">\(p_i\)</span> differ, and <span class="arithmatex">\(R\)</span> follows a Poisson-binomial distribution. Writing the probability of exactly <span class="arithmatex">\(j\)</span> presences as a sum over all subsets of <span class="arithmatex">\(j\)</span> genomes would be hopeless, since there are up to <span class="arithmatex">\(2^S\)</span> subsets. The distribution is instead obtained by adding the genomes one at a time. Let <span class="arithmatex">\(d_j\)</span> be the probability that exactly <span class="arithmatex">\(j\)</span> of the genomes processed so far contain the k-mer. Adding genome <span class="arithmatex">\(i\)</span> leaves <span class="arithmatex">\(j\)</span> unchanged if the genome does not contain the k-mer, and moves <span class="arithmatex">\(j-1\)</span> to <span class="arithmatex">\(j\)</span> if it does:</p>
<div class="arithmatex">\[
d'_j = d_j\,(1-p_i) + d_{j-1}\,p_i, \qquad d_0=1 \text{ before the first genome.}
\]</div>
<p>The table <span class="arithmatex">\(d\)</span> only needs to remember how many genomes were hit, not which ones, and this brings the cost down to <span class="arithmatex">\(O(S^2)\)</span>. The result does not depend on the order in which the genomes are processed. It does not depend on the k-mer either, so it is computed once per index. The upper tail is then</p>
<div class="arithmatex">\[
P_0(R\geq r) = 1-\sum_{j=0}^{r-1} d_j .
\]</div>
<p>Only <span class="arithmatex">\(d_0,\dots,d_{r-1}\)</span> are needed, so the cost is <span class="arithmatex">\(O(S\,r)\)</span>.</p>
<h3 id="approximation-with-equal-genome-sizes">Approximation with equal genome sizes</h3>
<p>If all genomes are given the same size, taken as the mean</p>
<div class="arithmatex">\[
\bar G=\frac1S\sum_{i=1}^{S}G_i, \qquad \bar p=\frac{2\bar G}{4^k},
\]</div>
<p>then <span class="arithmatex">\(R\)</span> follows a plain binomial distribution and the tail is</p>
<div class="arithmatex">\[
P_0(R\geq r)\simeq\sum_{j=r}^{S}\binom{S}{j}\,\bar p^{\,j}(1-\bar p)^{S-j}.
\]</div>
<p>This sum is the regularized incomplete beta function <span class="arithmatex">\(I_{\bar p}(r,\,S-r+1)\)</span>, available in standard statistical libraries. If it is summed by hand, the terms are obtained one from the next by <span class="arithmatex">\(t_{j+1}=t_j\,\frac{S-j}{j+1}\,\frac{\bar p}{1-\bar p}\)</span>.</p>
<p>A theorem of Hoeffding (Hoeffding 1956)<sup id="fnref2:Hoeffding1956-bx"><a class="footnote-ref" href="#fn:Hoeffding1956-bx">1</a></sup> shows that this approximation errs on the safe side. For a fixed mean, the Poisson-binomial distribution is more concentrated than the binomial one, so for <span class="arithmatex">\(r\geq S\bar p+1\)</span> the binomial tail is never smaller than the exact tail. Using the mean genome size therefore overestimates the chance of sharing and can only make the FDR more pessimistic. Gleser (Gleser 1975)<sup id="fnref:Gleser1975-yu"><a class="footnote-ref" href="#fn:Gleser1975-yu">8</a></sup> refines this result: replacing the <span class="arithmatex">\(p_i\)</span> by any partial average of themselves, not only by their full mean, also moves the upper tail towards the binomial one, provided <span class="arithmatex">\(r-1\geq S\bar p+2\)</span>. Grouping genomes of similar size is therefore safe in the same sense. What it loses is power when genome sizes differ by orders of magnitude, since one large genome then weighs as much as several small ones in <span class="arithmatex">\(\bar p\)</span>.</p>
<p>When all <span class="arithmatex">\(p_i\)</span> are small, which is the case for large <span class="arithmatex">\(k\)</span>, the number of hits is also well approximated by a Poisson distribution of mean <span class="arithmatex">\(\lambda=\sum_i p_i\)</span>, with an error bounded by <span class="arithmatex">\(\sum_i p_i^2\)</span> (Le Cam 1960)<sup id="fnref:Le-Cam1960-pe"><a class="footnote-ref" href="#fn:Le-Cam1960-pe">9</a></sup>.</p>
<p>For the numerical evaluation, the tail that does not contain the mode <span class="arithmatex">\(S\bar p\)</span> should be summed directly. When <span class="arithmatex">\(r&gt;S\bar p\)</span>, computing the sum from <span class="arithmatex">\(j=r\)</span> upwards is accurate even for very small probabilities, whereas computing <span class="arithmatex">\(1\)</span> minus the sum of the lower terms loses all precision once the tail falls below the machine epsilon.</p>
<h3 id="false-discovery-rate">False discovery rate</h3>
<p>The FDR of the call "present in at least <span class="arithmatex">\(r\)</span> genomes" is the fraction of retained k-mers expected to be pure chance:</p>
<div class="arithmatex">\[
\widehat{\mathrm{FDR}}(r)=\min\!\left(1,\ \frac{N\,P_0(R\geq r)}{n_{\geq r}}\right).
\]</div>
<p>The numerator is the number of k-mers expected to reach <span class="arithmatex">\(r\)</span> genomes by chance if the genomes were unrelated. Each of the <span class="arithmatex">\(N\)</span> canonical k-mers has probability <span class="arithmatex">\(P_0(R\geq r)\)</span> of doing so, and no further correction for multiple testing is needed because the number of tests is already in the product. The denominator <span class="arithmatex">\(n_{\geq r}\)</span> is the number of distinct k-mers actually observed in at least <span class="arithmatex">\(r\)</span> genomes, which is read directly from the index. The observed count mixes chance and true homology, so the ratio is an upper bound on the true FDR: it implicitly assumes that all k-mers are null, and it never underestimates the noise. The minimal <span class="arithmatex">\(r\)</span> for which <span class="arithmatex">\(\widehat{\mathrm{FDR}}(r)\)</span> falls below the accepted level, for example 5 %, is the sharing threshold to use.</p>
<p>As an illustration, take <span class="arithmatex">\(k=21\)</span> (so <span class="arithmatex">\(N=2.2\times10^{12}\)</span>), <span class="arithmatex">\(S=10\)</span> genomes and <span class="arithmatex">\(G_i=10^9\)</span> distinct canonical k-mers each, hence <span class="arithmatex">\(p=4.5\times10^{-4}\)</span>. The number of k-mers expected to be shared by chance is:</p>
<table>
<thead>
<tr>
<th style="text-align: right;"><span class="arithmatex">\(r\)</span></th>
<th style="text-align: right;"><span class="arithmatex">\(N\,P_0(R\geq r)\)</span></th>
</tr>
</thead>
<tbody>
<tr>
<td style="text-align: right;">2</td>
<td style="text-align: right;"><span class="arithmatex">\(2\times10^{7}\)</span></td>
</tr>
<tr>
<td style="text-align: right;">3</td>
<td style="text-align: right;"><span class="arithmatex">\(2.5\times10^{4}\)</span></td>
</tr>
<tr>
<td style="text-align: right;">4</td>
<td style="text-align: right;"><span class="arithmatex">\(\approx 20\)</span></td>
</tr>
<tr>
<td style="text-align: right;">5</td>
<td style="text-align: right;"><span class="arithmatex">\(\approx 0.01\)</span></td>
</tr>
</tbody>
</table>
<p>Chance alone produces millions of pairwise shared 21-mers at this genome size, a few tens of thousands shared by three genomes, and almost none shared by five or more. Whether a given <span class="arithmatex">\(r\)</span> is acceptable then depends on how many k-mers the index actually reports at that level.</p>
<h3 id="limits-of-the-model">Limits of the model</h3>
<ul>
<li>The FDR is computed per k-mer. Neighbouring k-mers of a shared region overlap and are not independent, so a single homologous region contributes many correlated k-mers. A per-region FDR would count regions instead of k-mers and is not covered here.</li>
<li>The probability <span class="arithmatex">\(p_i\)</span> assumes equiprobable, independent nucleotides. The composition biases discussed earlier in this page violate this assumption, and the resulting error is not accounted for.</li>
</ul>
<hr />
<p>Je n'ai rien modifié. Voici la formalisation des deux points, à valider avant de l'écrire dans <code>kmers.md</code>.</p>
<h2 id="1-fdr-dun-ensemble-de-k-mers-selectionnes">1. FDR d'un ensemble de k-mers sélectionnés</h2>
<p><strong>Estimation de l'espérance du nombre de faux positifs.</strong> Soit <span class="arithmatex">\(M\)</span> le nombre de k-mers canoniques testés, <span class="arithmatex">\(M = 4^k/2\)</span> pour k impair. Soit <span class="arithmatex">\(q(r)=P(X\ge r)\)</span> la queue du modèle nul. Le nombre attendu de k-mers présents par hasard dans au moins <span class="arithmatex">\(r\)</span> génomes vaut</p>
<div class="arithmatex">\[
E[\mathrm{FP}(r)] = M\,q(r).
\]</div>
<p>Cette égalité est exacte par linéarité de l'espérance, même si les k-mers chevauchants sont dépendants. La dépendance n'affecte que la variance, pas l'espérance. Seule l'approximation de Poisson par génome reste une approximation.</p>
<p><strong>FDR.</strong> Soit <span class="arithmatex">\(N(r)=\#\{i : r_i\ge r\}\)</span> le nombre observé de k-mers couvrant au moins <span class="arithmatex">\(r\)</span> génomes. Alors</p>
<div class="arithmatex">\[
\mathrm{FDR}(r)=\min\!\Big(1,\ \frac{M\,q(r)}{N(r)}\Big).
\]</div>
<p>Cette estimation est légèrement conservatrice. Elle suppose <span class="arithmatex">\(\pi_0=1\)</span>, c'est-à-dire que presque tout l'espace <span class="arithmatex">\(M\)</span> est nul, ce qui est vrai pour k grand.</p>
<p><strong>Ensemble hétérogène en <span class="arithmatex">\(r_i\)</span>.</strong> Si le jeu n'est pas défini par un seuil unique, on utilise le local fdr par classe de couverture :</p>
<div class="arithmatex">\[
\mathrm{fdr}(r)=\frac{M\,P(X=r)}{N_r},\qquad N_r=\#\{i : r_i=r\}.
\]</div>
<p>Le FDR de l'ensemble est alors la moyenne <span class="arithmatex">\(\frac1N\sum_i \mathrm{fdr}(r_i)\)</span>. Pour un ensemble défini par seuil <span class="arithmatex">\(r_{\min}\)</span>, cela redonne exactement la formule précédente.</p>
<p><strong>Remarque.</strong> <span class="arithmatex">\(M\,p\approx G\)</span>, d'où <span class="arithmatex">\(E[\mathrm{FP}(r)]\approx\binom{S}{r}\,G^r/M^{r-1}\)</span> en régime de petits <span class="arithmatex">\(p\)</span>. Cela donne une lecture directe de la quantité.</p>
<h2 id="2-markov-dordre-0-avec-frequences-inegales">2. Markov d'ordre 0 avec fréquences inégales</h2>
<p>Soit <span class="arithmatex">\(\pi_b\)</span> la fréquence du nucléotide <span class="arithmatex">\(b\)</span>. Pour un k-mer <span class="arithmatex">\(w\)</span> de composition <span class="arithmatex">\(n=(n_A,n_C,n_G,n_T)\)</span> avec <span class="arithmatex">\(\sum n_b=k\)</span>, on a <span class="arithmatex">\(P(w)=\prod_b \pi_b^{n_b}\)</span>.</p>
<p><strong>Attendu par génome.</strong> Le k-mer canonique <span class="arithmatex">\(K=\{w,\overline w\}\)</span> est vu quand le brin direct porte <span class="arithmatex">\(w\)</span> ou <span class="arithmatex">\(\overline w\)</span>, donc</p>
<div class="arithmatex">\[
\lambda_K = G\,\big(P(w)+P(\overline w)\big)=G\Big(\prod_b\pi_b^{n_b}+\pi_A^{n_T}\pi_T^{n_A}\pi_C^{n_G}\pi_G^{n_C}\Big).
\]</div>
<p>Cela ne dépend que de la composition. Si <span class="arithmatex">\(\pi_A=\pi_T\)</span> et <span class="arithmatex">\(\pi_C=\pi_G\)</span>, on retrouve <span class="arithmatex">\(2P(w)\)</span>.</p>
<p><strong>Faux positifs.</strong> Le test n'est plus homogène : chaque composition a sa propre <span class="arithmatex">\(p(n)=1-e^{-\lambda(n)}\)</span>. Il faut donc sommer sur les classes de composition (un k-mer et son complément inverse partagent le même <span class="arithmatex">\(\lambda\)</span>, d'où le facteur ½) :</p>
<div class="arithmatex">\[
E[\mathrm{FP}(r)]=\frac12\sum_{n}\binom{k}{n_A,n_C,n_G,n_T}\,P\big(X\ge r\mid p(n)\big).
\]</div>
<p>Cette somme porte sur <span class="arithmatex">\(\binom{k+3}{3}\)</span> compositions, soit 5 984 pour k = 31. Elle est donc légère avec la binomiale de la moyenne, ou avec une Poisson-binomiale par composition si les <span class="arithmatex">\(G_i\)</span> diffèrent. Si les <span class="arithmatex">\(\pi_b\)</span> varient d'un génome à l'autre (GC différent), les <span class="arithmatex">\(p_i(n)\)</span> se construisent par génome.</p>
<p><strong>Effet attendu.</strong> La moyenne de <span class="arithmatex">\(\lambda_w\)</span> sur tous les k-mers reste <span class="arithmatex">\(2G/4^k\)</span>, car <span class="arithmatex">\(\sum_w P(w)=1\)</span>. Dans le régime de petits <span class="arithmatex">\(p\)</span>, le terme dominant <span class="arithmatex">\(\binom{S}{r}p^r\)</span> est convexe en <span class="arithmatex">\(\lambda\)</span>. Par Jensen, le modèle équiprobable est donc le minimum de <span class="arithmatex">\(E[\mathrm{FP}]\)</span>, et la déviation de composition ne peut qu'augmenter le FDR. Les k-mers riches en AT ou GC dominent la somme, ce qui rejoint ton remarque sur la faible complexité. C'est un résultat sur le terme dominant, pas une preuve pour la queue exacte.</p>
<p><strong>Limites à signaler dans le texte.</strong> Markov 0 ignore les dinucléotides, comme les CpG ou les répétitions, et le <span class="arithmatex">\(M\)</span> du modèle nul reste celui de l'espace total.</p>
<p>Veux-tu que j'écrive ces deux sections, avec les corrections de numérique évoquées plus haut (<code>expm1</code>, somme directe de la queue) ?</p>
<div class="footnote">
<hr />
<ol>
<li id="fn:Hoeffding1956-bx">
<p>Hoeffding W (1956) <a href="https://doi.org/10.1214/aoms/1177728178">[On the distribution of the number of successes in independent trials]{.nocase}</a>. <em>The Annals of Mathematical Statistics</em>, <strong>27</strong>, 713--721.&#160;<a class="footnote-backref" href="#fnref:Hoeffding1956-bx" title="Jump back to footnote 1 in the text">&#8617;</a><a class="footnote-backref" href="#fnref2:Hoeffding1956-bx" title="Jump back to footnote 1 in the text">&#8617;</a></p>
</li>
<li id="fn:Fraser1995-jc">
<p>Fraser CM, Gocayne JD, White O, Adams MD, Clayton RA, Fleischmann RD, Bult CJ, Kerlavage AR, Sutton G, Kelley JM, Fritchman RD, Weidman JF, Small KV, Sandusky M, Fuhrmann J, Nguyen D, Utterback TR, Saudek DM, Phillips CA, Merrick JM, Tomb JF, Dougherty BA, Bott KF, Hu PC, Lucier TS, Peterson SN, Smith HO, Hutchison CA 3rd, Venter JC (1995) <a href="https://doi.org/10.1126/science.270.5235.397">The minimal gene complement of mycoplasma genitalium</a>. <em>Science (New York, N.Y.)</em>, <strong>270</strong>, 397--403.&#160;<a class="footnote-backref" href="#fnref:Fraser1995-jc" title="Jump back to footnote 2 in the text">&#8617;</a></p>
</li>
<li id="fn:Schneiker2007-gs">
<p>Schneiker S, Perlova O, Kaiser O, Gerth K, Alici A, Altmeyer MO, Bartels D, Bekel T, Beyer S, Bode E, Bode HB, Bolten CJ, Choudhuri JV, Doss S, Elnakady YA, Frank B, Gaigalat L, Goesmann A, Groeger C, Gross F, Jelsbak L, Jelsbak L, Kalinowski J, Kegler C, Knauber T, Konietzny S, Kopp M, Krause L, Krug D, Linke B, Mahmud T, Martinez-Arias R, McHardy AC, Merai M, Meyer F, Mormann S, Muñoz-Dorado J, Perez J, Pradella S, Rachid S, Raddatz G, Rosenau F, Rückert C, Sasse F, Scharfe M, Schuster SC, Suen G, Treuner-Lange A, Velicer GJ, Vorhölter F-J, Weissman KJ, Welch RD, Wenzel SC, Whitworth DE, Wilhelm S, Wittmann C, Blöcker H, Pühler A, Müller R (2007) <a href="https://doi.org/10.1038/nbt1354">Complete genome sequence of the myxobacterium sorangium cellulosum</a>. <em>Nature Biotechnology</em>, <strong>25</strong>, 1281--1289.&#160;<a class="footnote-backref" href="#fnref:Schneiker2007-gs" title="Jump back to footnote 3 in the text">&#8617;</a></p>
</li>
<li id="fn:Fleischmann2014-ed">
<p>Fleischmann A, Michael TP, Rivadavia F, Sousa A, Wang W, Temsch EM, Greilhuber J, Müller KF, Heubl G (2014) <a href="https://doi.org/10.1093/aob/mcu189">Evolution of genome size and chromosome number in the carnivorous plant genus genlisea (lentibulariaceae), with a new estimate of the minimum genome size in angiosperms</a>. <em>Annals of Botany</em>, <strong>114</strong>, 1651--1663.&#160;<a class="footnote-backref" href="#fnref:Fleischmann2014-ed" title="Jump back to footnote 4 in the text">&#8617;</a></p>
</li>
<li id="fn:Fernandez2024-kd">
<p>Fernández P, Amice R, Bruy D, Christenhusz MJM, Leitch IJ, Leitch AL, Pokorny L, Hidalgo O, Pellicer J (2024) <a href="https://doi.org/10.1016/j.isci.2024.109889">A 160 gbp fork fern genome shatters size record for eukaryotes</a>. <em>iScience</em>, <strong>27</strong>, 109889.&#160;<a class="footnote-backref" href="#fnref:Fernandez2024-kd" title="Jump back to footnote 5 in the text">&#8617;</a></p>
</li>
<li id="fn:Karlin1988">
<p>Karlin S, Ost F (1988) Maximal length of common words among random letter sequences. <em>The Annals of Probability</em>, <strong>16</strong>, 535--563.&#160;<a class="footnote-backref" href="#fnref:Karlin1988" title="Jump back to footnote 6 in the text">&#8617;</a></p>
</li>
<li id="fn:Karlin1985">
<p>Karlin S, Ost F (1985) Maximal segmental match length among random sequences from a finite alphabet. In: <em>Proceedings of the berkeley conference in honor of jerzy neyman and jack kiefer</em> (eds Le Cam LM, Olshen RA), pp. 225--243. Wadsworth, Monterey, CA.&#160;<a class="footnote-backref" href="#fnref:Karlin1985" title="Jump back to footnote 7 in the text">&#8617;</a></p>
</li>
<li id="fn:Gleser1975-yu">
<p>Gleser LJ (1975) <a href="https://doi.org/10.1214/aop/1176996461">[On the distribution of the number of successes in independent trials]{.nocase}</a>. <em>Annals of Probability</em>, <strong>3</strong>, 182--188.&#160;<a class="footnote-backref" href="#fnref:Gleser1975-yu" title="Jump back to footnote 8 in the text">&#8617;</a></p>
</li>
<li id="fn:Le-Cam1960-pe">
<p>Le Cam L (1960) <a href="https://doi.org/10.2140/pjm.1960.10.1181">[An approximation theorem for the Poisson binomial distribution]{.nocase}</a>. <em>Pacific Journal of Mathematics</em>, <strong>10</strong>, 1181--1197.&#160;<a class="footnote-backref" href="#fnref:Le-Cam1960-pe" title="Jump back to footnote 9 in the text">&#8617;</a></p>
</li>
</ol>
</div>
</article>
</div>
<script>var target=document.getElementById(location.hash.slice(1));target&&target.name&&(target.checked=target.name.startsWith("__tabbed_"))</script>
</div>
</main>
<footer class="md-footer">
<div class="md-footer-meta md-typeset">
<div class="md-footer-meta__inner md-grid">
<div class="md-copyright">
Made with
<a href="https://squidfunk.github.io/mkdocs-material/" target="_blank" rel="noopener">
Material for MkDocs
</a>
</div>
</div>
</div>
</footer>
</div>
<div class="md-dialog" data-md-component="dialog">
<div class="md-dialog__inner md-typeset"></div>
</div>
<script id="__config" type="application/json">{"annotate": null, "base": "../../..", "features": [], "search": "../../../assets/javascripts/workers/search.2c215733.min.js", "tags": null, "translations": {"clipboard.copied": "Copied to clipboard", "clipboard.copy": "Copy to clipboard", "search.result.more.one": "1 more on this page", "search.result.more.other": "# more on this page", "search.result.none": "No matching documents", "search.result.one": "1 matching document", "search.result.other": "# matching documents", "search.result.placeholder": "Type to start searching", "search.result.term.missing": "Missing", "select.version": "Select version"}, "version": null}</script>
<script src="../../../assets/javascripts/bundle.79ae519e.min.js"></script>
<script src="https://unpkg.com/mathjax@3/es5/tex-mml-chtml.js"></script>
</body>
</html>