From cf5b4baa54eda0e340af9325c3149ccb421e2c01 Mon Sep 17 00:00:00 2001
From: Eric Coissac <eric.coissac@metabarcoding.org>
Date: Mon, 30 May 2022 16:28:59 +0200
Subject: [PATCH] Small bug in sequence counting on merge

---
 pkg/obichunk/subchunks.go       |  2 +-
 pkg/obichunk/unique.go          |  7 ++++++-
 pkg/obiseq/merge.go             |  3 ++-
 pkg/obitools/obiuniq/options.go |  2 +-
 pkg/obitools/obiuniq/unique.go  | 11 +++++++++++
 5 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/pkg/obichunk/subchunks.go b/pkg/obichunk/subchunks.go
index dbce3d6..dc40edf 100644
--- a/pkg/obichunk/subchunks.go
+++ b/pkg/obichunk/subchunks.go
@@ -11,7 +11,7 @@ import (
 )
 
 //
-// Interface for sorting a list of sequences accoording to
+// Interface for sorting a list of sequences according to
 // their classes
 //
 
diff --git a/pkg/obichunk/unique.go b/pkg/obichunk/unique.go
index 33c3e65..ddfddb6 100644
--- a/pkg/obichunk/unique.go
+++ b/pkg/obichunk/unique.go
@@ -9,6 +9,9 @@ import (
 	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
 )
 
+// Runs dereplication algorithm on a  obiiter.IBioSequenceBatch
+// iterator.  
+
 func IUniqueSequence(iterator obiiter.IBioSequenceBatch,
 	options ...WithOption) (obiiter.IBioSequenceBatch, error) {
 
@@ -62,7 +65,9 @@ func IUniqueSequence(iterator obiiter.IBioSequenceBatch,
 		return neworder
 	}
 
-	var ff func(obiiter.IBioSequenceBatch, *obiseq.BioSequenceClassifier, int)
+	var ff func(obiiter.IBioSequenceBatch,
+		*obiseq.BioSequenceClassifier,
+		int)
 
 	cat := opts.Categories()
 	na := opts.NAValue()
diff --git a/pkg/obiseq/merge.go b/pkg/obiseq/merge.go
index 29bb581..7737ef3 100644
--- a/pkg/obiseq/merge.go
+++ b/pkg/obiseq/merge.go
@@ -37,6 +37,7 @@ func (sequence *BioSequence) StatsOn(key string, na string) StatsOnValues {
 			newstat = false
 		case map[string]interface{}:
 			stats = make(StatsOnValues, len(istat))
+			newstat = false
 			var err error
 			for k, v := range istat {
 				stats[k], err = goutils.InterfaceToInt(v)
@@ -161,7 +162,7 @@ func (sequences BioSequenceSlice) Merge(na string, statsOn []string) *BioSequenc
 	seq.SetQualities(nil)
 
 	if len(sequences) == 1 {
-		seq.Annotations()["count"] = 1
+		seq.Annotations()["count"] = seq.Count()
 		for _, v := range statsOn {
 			seq.StatsOn(v, na)
 		}
diff --git a/pkg/obitools/obiuniq/options.go b/pkg/obitools/obiuniq/options.go
index 7caa82b..2b64718 100644
--- a/pkg/obitools/obiuniq/options.go
+++ b/pkg/obitools/obiuniq/options.go
@@ -41,7 +41,7 @@ func UniqueOptionSet(options *getoptions.GetOpt) {
 }
 
 // OptionSet adds to the basic option set every options declared for
-// the obipcr command
+// the obiuniq command
 func OptionSet(options *getoptions.GetOpt) {
 	obiconvert.OptionSet(options)
 	UniqueOptionSet(options)
diff --git a/pkg/obitools/obiuniq/unique.go b/pkg/obitools/obiuniq/unique.go
index c3d6b77..356d9b7 100644
--- a/pkg/obitools/obiuniq/unique.go
+++ b/pkg/obitools/obiuniq/unique.go
@@ -16,6 +16,11 @@ func Unique(sequences obiiter.IBioSequenceBatch) obiiter.IBioSequenceBatch {
 		obichunk.OptionBatchCount(CLINumberOfChunks()),
 	)
 
+	//
+	// Considers if data splitting must be done on disk or in memory
+	//
+	// --on-disk command line option
+
 	if CLIUniqueInMemory() {
 		log.Printf("Running dereplication in memory on %d chunks", CLINumberOfChunks())
 		options = append(options, obichunk.OptionSortOnMemory())
@@ -24,6 +29,12 @@ func Unique(sequences obiiter.IBioSequenceBatch) obiiter.IBioSequenceBatch {
 		options = append(options, obichunk.OptionSortOnDisk())
 	}
 
+	//
+	// Considers if sequences observed a singletime in the dataset have to
+	// be conserved in the output
+	//
+	// --no-singleton
+
 	if CLINoSingleton() {
 		log.Printf("Removing sigletons from the output")
 		options = append(options, obichunk.OptionsNoSingleton())