add the --min-sample-count option to obiclean.

pkg/obialign/fastlcs.go

@@ -21,15 +21,15 @@ func encodeValues(score, length int, out bool) uint64 {
 	return fo
 }
 
-func _isout(value uint64) bool {
-	const outmask = uint64(1) << dwsize
-	return (value & outmask) == 0
-}
+// func _isout(value uint64) bool {
+// 	const outmask = uint64(1) << dwsize
+// 	return (value & outmask) == 0
+// }
 
-func _lpath(value uint64) int {
-	const mask = uint64(1<<wsize) - 1
-	return int(((value + 1) ^ mask) & mask)
-}
+// func _lpath(value uint64) int {
+// 	const mask = uint64(1<<wsize) - 1
+// 	return int(((value + 1) ^ mask) & mask)
+// }
 
 func decodeValues(value uint64) (int, int, bool) {
 	const mask = uint64(1<<wsize) - 1
@@ -57,4 +57,3 @@ func _setout(value uint64) uint64 {
 var _empty = encodeValues(0, 0, false)
 var _out = encodeValues(0, 30000, true)
 var _notavail = encodeValues(0, 30000, false)
-

pkg/obioptions/version.go

@@ -8,7 +8,7 @@ import (
 // corresponds to the last commit, and not the one when the file will be
 // commited
 
-var _Commit = "fb6f857"
+var _Commit = "51d11aa"
 var _Version = "Release 4.2.0"
 
 // Version returns the version of the obitools package.

pkg/obiseq/predicate.go

@@ -196,6 +196,16 @@ func IsShorterOrEqualTo(length int) SequencePredicate {
 	return f
 }
 
+func OccurInAtleast(sample string, n int) SequencePredicate {
+	desc := MakeStatsOnDescription(sample)
+	f := func(sequence *BioSequence) bool {
+		stats := sequence.StatsOn(desc, "NA")
+		return len(stats) >= n
+	}
+
+	return f
+}
+
 func IsSequenceMatch(pattern string) SequencePredicate {
 	pat, err := regexp.Compile("(?i)" + pattern)
 

pkg/obitools/obiclean/obiclean.go

@@ -368,7 +368,12 @@ func CLIOBIClean(itertator obiiter.IBioSequence) obiiter.IBioSequence {
 	iter := annotateOBIClean(source, db, samples, SampleAttribute(), "NA")
 
 	if OnlyHead() {
-		iter = iter.FilterOn(IsHead, 1000)
+		iter = iter.FilterOn(IsHead, obidefault.BatchSize())
+	}
+
+	if MinSampleCount() > 1 {
+		sc := obiseq.OccurInAtleast(SampleAttribute(), MinSampleCount())
+		iter = iter.FilterOn(sc, obidefault.BatchSize())
 	}
 
 	return iter

pkg/obitools/obiclean/options.go

@@ -16,6 +16,7 @@ var _onlyHead = false
 
 var _saveGraph = "__@@NOSAVE@@__"
 var _saveRatio = "__@@NOSAVE@@__"
+var _minSample = 1
 
 func ObicleanOptionSet(options *getoptions.GetOpt) {
 	options.StringVar(&_sampleAttribute, "sample", _sampleAttribute,
@@ -55,6 +56,9 @@ func ObicleanOptionSet(options *getoptions.GetOpt) {
 			"The ratio file follows the csv format."),
 	)
 
+	options.IntVar(&_minSample, "min-sample-count", _minSample,
+		options.Description("Minimum number of samples a sequence must be present in to be considered in the analysis."),
+	)
 }
 
 func OptionSet(options *getoptions.GetOpt) {
@@ -111,3 +115,8 @@ func IsSaveRatioTable() bool {
 func RatioTableFilename() string {
 	return _saveRatio
 }
+
+// It returns the minimum number of samples a sequence must be present in to be considered in the analysis
+func MinSampleCount() int {
+	return _minSample
+}

pkg/obitools/obipairing/pairing.go

@@ -238,7 +238,7 @@ func IAssemblePESequencesBatch(iterator obiiter.IBioSequence,
 		log.Printf("End of the sequence Pairing")
 	}()
 
-	f := func(iterator obiiter.IBioSequence, wid int) {
+	f := func(iterator obiiter.IBioSequence) {
 		arena := obialign.MakePEAlignArena(150, 150)
 		shifts := make(map[int]int)
 
@@ -263,9 +263,9 @@ func IAssemblePESequencesBatch(iterator obiiter.IBioSequence,
 	log.Printf("Start of the sequence Pairing using %d workers\n", nworkers)
 
 	for i := 0; i < nworkers-1; i++ {
-		go f(iterator.Split(), i)
+		go f(iterator.Split())
 	}
-	go f(iterator, nworkers-1)
+	go f(iterator)
 	return newIter
 
 }