mirror of
https://github.com/metabarcoding/obitools4.git
synced 2025-12-12 18:30:25 +00:00
Add obiminion first version
Former-commit-id: aa5ace7bd4d2266333715fca7094d1c3cbbb5e6d
This commit is contained in:
Eric Coissac
@@ -4,92 +4,93 @@ import (
|
||||
"fmt"
|
||||
"os"
|
||||
"path"
|
||||
"sort"
|
||||
"slices"
|
||||
|
||||
log "github.com/sirupsen/logrus"
|
||||
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiformats"
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiiter"
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obikmer"
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiseq"
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obisuffix"
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiutils"
|
||||
)
|
||||
|
||||
func BuildConsensus(seqs obiseq.BioSequenceSlice,
|
||||
consensus_id string,
|
||||
kmer_size int, quorum float64,
|
||||
min_depth float64,
|
||||
max_length int,
|
||||
save_graph bool, dirname string) (*obiseq.BioSequence, error) {
|
||||
|
||||
if save_graph {
|
||||
if dirname == "" {
|
||||
dirname = "."
|
||||
}
|
||||
|
||||
if stat, err := os.Stat(dirname); err != nil || !stat.IsDir() {
|
||||
// path does not exist or is not directory
|
||||
os.RemoveAll(dirname)
|
||||
err := os.Mkdir(dirname, 0755)
|
||||
|
||||
if err != nil {
|
||||
log.Panicf("Cannot create directory %s for saving graphs", dirname)
|
||||
}
|
||||
}
|
||||
|
||||
fasta, err := os.Create(path.Join(dirname, fmt.Sprintf("%s.fasta", consensus_id)))
|
||||
|
||||
if err == nil {
|
||||
defer fasta.Close()
|
||||
fasta.Write(obiformats.FormatFastaBatch(obiiter.MakeBioSequenceBatch(0, seqs), obiformats.FormatFastSeqJsonHeader, false))
|
||||
fasta.Close()
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
log.Printf("Number of reads : %d\n", len(seqs))
|
||||
|
||||
if kmer_size < 0 {
|
||||
longest := make([]int, len(seqs))
|
||||
|
||||
for i := range seqs {
|
||||
s := seqs[i : i+1]
|
||||
for i, seq := range seqs {
|
||||
s := obiseq.BioSequenceSlice{seq}
|
||||
sa := obisuffix.BuildSuffixArray(&s)
|
||||
longest[i] = obiutils.MaxSlice(sa.CommonSuffix())
|
||||
longest[i] = slices.Max(sa.CommonSuffix())
|
||||
}
|
||||
|
||||
o := obiutils.Order(sort.IntSlice(longest))
|
||||
i := int(float64(len(seqs)) * quorum)
|
||||
// o := obiutils.Order(sort.IntSlice(longest))
|
||||
// i := int(float64(len(seqs)) * quorum)
|
||||
|
||||
kmer_size = longest[o[i]] + 1
|
||||
// if i >= len(o) {
|
||||
// i = len(o) - 1
|
||||
// }
|
||||
|
||||
kmer_size = slices.Max(longest) + 1
|
||||
|
||||
// kmer_size = longest[o[i]] + 1
|
||||
log.Printf("estimated kmer size : %d", kmer_size)
|
||||
}
|
||||
|
||||
graph := obikmer.MakeDeBruijnGraph(kmer_size)
|
||||
var graph *obikmer.DeBruijnGraph
|
||||
for {
|
||||
graph = obikmer.MakeDeBruijnGraph(kmer_size)
|
||||
|
||||
for _, s := range seqs {
|
||||
graph.Push(s)
|
||||
for _, s := range seqs {
|
||||
graph.Push(s)
|
||||
}
|
||||
|
||||
if !graph.HasCycle() {
|
||||
break
|
||||
}
|
||||
|
||||
kmer_size++
|
||||
log.Infof("Cycle detected, increasing kmer size to %d\n", kmer_size)
|
||||
}
|
||||
|
||||
log.Printf("Graph size : %d\n", graph.Len())
|
||||
total_kmer := graph.Len()
|
||||
|
||||
threshold := 0
|
||||
|
||||
switch {
|
||||
case min_depth < 0:
|
||||
spectrum := graph.LinkSpectrum()
|
||||
cum := make(map[int]int)
|
||||
|
||||
spectrum[1] = 0
|
||||
for i := 2; i < len(spectrum); i++ {
|
||||
spectrum[i] += spectrum[i-1]
|
||||
cum[spectrum[i]]++
|
||||
}
|
||||
|
||||
max := 0
|
||||
kmax := 0
|
||||
for k, obs := range cum {
|
||||
if obs > max {
|
||||
max = obs
|
||||
kmax = k
|
||||
}
|
||||
}
|
||||
|
||||
for i, total := range spectrum {
|
||||
if total == kmax {
|
||||
threshold = i
|
||||
break
|
||||
}
|
||||
}
|
||||
threshold /= 2
|
||||
case min_depth >= 1:
|
||||
threshold = int(min_depth)
|
||||
default:
|
||||
threshold = int(float64(len(seqs)) * min_depth)
|
||||
}
|
||||
|
||||
graph.FilterMin(threshold)
|
||||
|
||||
log.Printf("Graph size : %d\n", graph.Len())
|
||||
|
||||
if save_graph {
|
||||
|
||||
file, err := os.Create(path.Join(dirname,
|
||||
fmt.Sprintf("%s.gml", seqs[0].Source())))
|
||||
fmt.Sprintf("%s_raw_consensus.gml", consensus_id)))
|
||||
|
||||
if err != nil {
|
||||
fmt.Println(err)
|
||||
@@ -99,29 +100,133 @@ func BuildConsensus(seqs obiseq.BioSequenceSlice,
|
||||
}
|
||||
}
|
||||
|
||||
id := seqs[0].Source()
|
||||
if id == "" {
|
||||
id = seqs[0].Id()
|
||||
}
|
||||
seq, err := graph.LongestConsensus(id)
|
||||
log.Printf("Graph size : %d\n", graph.Len())
|
||||
total_kmer := graph.Len()
|
||||
|
||||
// threshold := 0
|
||||
|
||||
// switch {
|
||||
// case min_depth < 0:
|
||||
// spectrum := graph.WeightSpectrum()
|
||||
// cum := make(map[int]int)
|
||||
|
||||
// spectrum[1] = 0
|
||||
// for i := 2; i < len(spectrum); i++ {
|
||||
// spectrum[i] += spectrum[i-1]
|
||||
// cum[spectrum[i]]++
|
||||
// }
|
||||
|
||||
// max := 0
|
||||
// kmax := 0
|
||||
// for k, obs := range cum {
|
||||
// if obs > max {
|
||||
// max = obs
|
||||
// kmax = k
|
||||
// }
|
||||
// }
|
||||
|
||||
// for i, total := range spectrum {
|
||||
// if total == kmax {
|
||||
// threshold = i
|
||||
// break
|
||||
// }
|
||||
// }
|
||||
// threshold /= 2
|
||||
|
||||
// if threshold < 1 {
|
||||
// threshold = 1
|
||||
// }
|
||||
|
||||
// log.Info("Estimated kmer_min_occur = ", threshold)
|
||||
// case min_depth >= 1:
|
||||
// threshold = int(min_depth)
|
||||
// default:
|
||||
// threshold = int(float64(len(seqs)) * min_depth)
|
||||
// }
|
||||
|
||||
// graph.FilterMinWeight(threshold)
|
||||
|
||||
// log.Printf("Graph size : %d\n", graph.Len())
|
||||
|
||||
// if save_graph {
|
||||
|
||||
// file, err := os.Create(path.Join(dirname,
|
||||
// fmt.Sprintf("%s_consensus.gml", consensus_id)))
|
||||
|
||||
// if err != nil {
|
||||
// fmt.Println(err)
|
||||
// } else {
|
||||
// file.WriteString(graph.Gml())
|
||||
// file.Close()
|
||||
// }
|
||||
// }
|
||||
|
||||
seq, err := graph.LongestConsensus(consensus_id, max_length)
|
||||
|
||||
sumCount := 0
|
||||
|
||||
for _, s := range seqs {
|
||||
sumCount += s.Count()
|
||||
if seq != nil {
|
||||
for _, s := range seqs {
|
||||
sumCount += s.Count()
|
||||
}
|
||||
|
||||
seq.SetCount(sumCount)
|
||||
seq.SetAttribute("seq_length", seq.Len())
|
||||
seq.SetAttribute("kmer_size", kmer_size)
|
||||
//seq.SetAttribute("kmer_min_occur", threshold)
|
||||
seq.SetAttribute("kmer_max_occur", graph.MaxWeight())
|
||||
seq.SetAttribute("filtered_graph_size", graph.Len())
|
||||
seq.SetAttribute("full_graph_size", total_kmer)
|
||||
}
|
||||
|
||||
seq.SetCount(sumCount)
|
||||
seq.SetAttribute("seq_length", seq.Len())
|
||||
seq.SetAttribute("kmer_size", kmer_size)
|
||||
seq.SetAttribute("kmer_min_occur", threshold)
|
||||
seq.SetAttribute("kmer_max_occur", graph.MaxLink())
|
||||
seq.SetAttribute("filtered_graph_size", graph.Len())
|
||||
seq.SetAttribute("full_graph_size", total_kmer)
|
||||
|
||||
return seq, err
|
||||
}
|
||||
|
||||
// func BuildConsensusWithTimeout(seqs obiseq.BioSequenceSlice,
|
||||
// kmer_size int, quorum float64,
|
||||
// min_depth float64,
|
||||
// save_graph bool, dirname string, timeout time.Duration) (*obiseq.BioSequence, error) {
|
||||
|
||||
// ctx, cancel := context.WithTimeout(context.Background(), timeout)
|
||||
// defer cancel()
|
||||
|
||||
// consensus := func() *obiseq.BioSequence {
|
||||
// cons, err := BuildConsensus(seqs, kmer_size, quorum, min_depth, save_graph, dirname,)
|
||||
// if err != nil {
|
||||
// cons = nil
|
||||
// }
|
||||
|
||||
// return cons
|
||||
// }
|
||||
|
||||
// computation := func() <-chan *obiseq.BioSequence {
|
||||
// result := make(chan *obiseq.BioSequence)
|
||||
|
||||
// go func() {
|
||||
// select {
|
||||
// case <-ctx.Done():
|
||||
// result <- nil
|
||||
// default:
|
||||
// result <- consensus()
|
||||
|
||||
// }
|
||||
// }()
|
||||
|
||||
// return result
|
||||
// }
|
||||
|
||||
// calcResult := computation()
|
||||
|
||||
// select {
|
||||
// case result := <-calcResult:
|
||||
// if result == nil {
|
||||
// return nil, fmt.Errorf("cannot compute consensus")
|
||||
// }
|
||||
// return result, nil
|
||||
// case <-ctx.Done():
|
||||
// return nil, fmt.Errorf("compute consensus timeout, exiting")
|
||||
// }
|
||||
// }
|
||||
|
||||
func Consensus(iterator obiiter.IBioSequence) obiiter.IBioSequence {
|
||||
newIter := obiiter.MakeIBioSequence()
|
||||
size := 10
|
||||
@@ -153,10 +258,19 @@ func Consensus(iterator obiiter.IBioSequence) obiiter.IBioSequence {
|
||||
for iterator.Next() {
|
||||
seqs := iterator.Get()
|
||||
|
||||
consensus, err := BuildConsensus(seqs.Slice(),
|
||||
sequences := seqs.Slice()
|
||||
|
||||
id := sequences[0].Source()
|
||||
if id == "" {
|
||||
id = sequences[0].Id()
|
||||
}
|
||||
consensus, err := BuildConsensus(sequences,
|
||||
id,
|
||||
CLIKmerSize(), CLIThreshold(),
|
||||
CLIKmerDepth(),
|
||||
CLISaveGraphToFiles(), CLIGraphFilesDirectory(),
|
||||
CLIMaxConsensusLength(),
|
||||
CLISaveGraphToFiles(),
|
||||
CLIGraphFilesDirectory(),
|
||||
)
|
||||
|
||||
if err == nil {
|
||||
|
||||
@@ -9,6 +9,7 @@ var _saveGraph = "__@@NOSAVE@@__"
|
||||
var _kmerSize = -1
|
||||
var _threshold = 0.99
|
||||
var _mindepth = -1.0
|
||||
var _consensus_max_length = -1
|
||||
|
||||
func ObiconsensusOptionSet(options *getoptions.GetOpt) {
|
||||
|
||||
@@ -38,6 +39,12 @@ func ObiconsensusOptionSet(options *getoptions.GetOpt) {
|
||||
"Default value = -1, which means that the DEPTH is estimated from the data"),
|
||||
)
|
||||
|
||||
options.IntVar(&_consensus_max_length, "consensus-max-length", _consensus_max_length,
|
||||
options.ArgName("LENGTH"),
|
||||
options.Description("Maximum length of the consensus sequence. "+
|
||||
"Default value = -1, which means that no limit is applied"),
|
||||
)
|
||||
|
||||
}
|
||||
|
||||
func OptionSet(options *getoptions.GetOpt) {
|
||||
@@ -67,3 +74,7 @@ func CLIKmerDepth() float64 {
|
||||
func CLIThreshold() float64 {
|
||||
return _threshold
|
||||
}
|
||||
|
||||
func CLIMaxConsensusLength() int {
|
||||
return _consensus_max_length
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user