Force sequence reading to produce lowercase sequences.

Adds two columns to the obiclean ratio csv file

pkg/obitools/obiclean/graph.go

@@ -18,6 +18,8 @@ import (
 
 type Ratio struct {
 	Sample string
+	SeqID  string
+	status string
 	From   int
 	To     int
 	CFrom  int
@@ -97,12 +99,14 @@ func EmpiricalDistCsv(filename string, data [][]Ratio) {
 
 	bar := progressbar.NewOptions(len(data), pbopt...)
 
-	fmt.Fprintln(file, "Sample,From,To,Weight_from,Weight_to,Count_from,Count_to,Position,length")
+	fmt.Fprintln(file, "Sample,Father_id,Father_status,From,To,Weight_from,Weight_to,Count_from,Count_to,Position,length")
 	for code, dist := range data {
 		a1, a2 := intToNucPair(code)
 		for _, ratio := range dist {
-			fmt.Fprintf(file, "%s,%c,%c,%d,%d,%d,%d,%d,%d\n",
+			fmt.Fprintf(file, "%s,%s,%s,%c,%c,%d,%d,%d,%d,%d,%d\n",
 				ratio.Sample,
+				ratio.SeqID,
+				ratio.status,
 				a1, a2,
 				ratio.From,
 				ratio.To,
@@ -463,7 +467,13 @@ func EstimateRatio(samples map[string]*[]*seqPCR, minStatRatio int) [][]Ratio {
 			for _, edge := range seq.Edges {
 				father := (*seqs)[edge.Father]
 				if father.Weight >= minStatRatio && edge.Dist == 1 {
-					ratio[edge.NucPair] = append(ratio[edge.NucPair], Ratio{name, father.Weight, seq.Weight, father.Count, seq.Count, edge.Pos, father.Sequence.Len()})
+					ratio[edge.NucPair] = append(ratio[edge.NucPair],
+						Ratio{name,
+							father.Sequence.Id(), Status(father.Sequence)[name],
+							father.Weight, seq.Weight,
+							father.Count, seq.Count,
+							edge.Pos,
+							father.Sequence.Len()})
 				}
 			}
 

pkg/obitools/obiclean/obiclean.go

@@ -19,6 +19,7 @@ type seqPCR struct {
 	SonCount  int
 	AddedSons int
 	Edges     []Edge
+	Cluster   map[int]bool        // used as the set of head sequences associated to that sequence 
 }
 
 // buildSamples sorts the sequences by samples
@@ -183,13 +184,53 @@ func GetMutation(sequence *obiseq.BioSequence) map[string]string {
 	return mutation
 }
 
+func GetCluster(sequence *obiseq.BioSequence) map[string]string {
+	annotation := sequence.Annotations()
+	icluster, ok := annotation["obiclean_cluster"]
+	var cluster map[string]string
+
+	if ok {
+		switch icluster := icluster.(type) {
+		case map[string]string:
+			cluster = icluster
+		case map[string]interface{}:
+			cluster = make(map[string]string)
+			for k, v := range icluster {
+				cluster[k] = fmt.Sprint(v)
+			}
+		}
+	} else {
+		cluster = make(map[string]string)
+		annotation["obiclean_cluster"] = cluster
+	}
+
+	return cluster
+}
+
+
+// func Cluster(sample map[string]*([]*seqPCR)) {
+// 	for _, graph := range sample {
+// 		for _, s := range *graph {
+// 			cluster := GetCluster(s.Sequence)
+// 			if len(s.Edges) > 0 {
+// 				for _, f := range s.Edges {
+
+// 				}	
+// 			} else {
+// 				cluster
+// 			}
+
+// 		}
+// 	}
+// }
+
 func Mutation(sample map[string]*([]*seqPCR)) {
 	for _, graph := range sample {
 		for _, s := range *graph {
 			for _, f := range s.Edges {
 				id := (*graph)[f.Father].Sequence.Id()
 				GetMutation(s.Sequence)[id] = fmt.Sprintf("(%c)->(%c)@%d",
-					f.From, f.To, f.Pos + 1)
+					f.From, f.To, f.Pos+1)
 			}
 		}
 	}
@@ -277,14 +318,6 @@ func IOBIClean(itertator obiiter.IBioSequenceBatch) obiiter.IBioSequenceBatch {
 		}
 	}
 
-	if IsSaveRatioTable() {
-		all_ratio := EstimateRatio(samples, MinCountToEvalMutationRate())
-		EmpiricalDistCsv(RatioTableFilename(), all_ratio)
-	}
-
-	if SaveGraphToFiles() {
-		SaveGMLGraphs(GraphFilesDirectory(), samples, MinCountToEvalMutationRate())
-	}
 
 	Mutation(samples)
 
@@ -310,6 +343,16 @@ func IOBIClean(itertator obiiter.IBioSequenceBatch) obiiter.IBioSequenceBatch {
 		bar.Add(1)
 	}
 
+	if SaveGraphToFiles() {
+		SaveGMLGraphs(GraphFilesDirectory(), samples, MinCountToEvalMutationRate())
+	}
+
+	if IsSaveRatioTable() {
+		all_ratio := EstimateRatio(samples, MinCountToEvalMutationRate())
+		EmpiricalDistCsv(RatioTableFilename(), all_ratio)
+	}
+
+
 	iter := annotateOBIClean(db, samples, SampleAttribute(), "NA")
 
 	if OnlyHead() {

pkg/obitools/obimultiplex/demultiplex.go

@@ -56,5 +56,5 @@ func IExtractBarcode(iterator obiiter.IBioSequenceBatch) (obiiter.IBioSequenceBa
 	}
 	log.Printf("Sequence demultiplexing using %d workers\n", obioptions.CLIParallelWorkers())
 
-	return newIter, nil
+	return newIter.Speed("Demultiplexing"), nil
 }

pkg/obitools/obitag/obitag.go

@@ -66,7 +66,9 @@ func FindClosests(sequence *obiseq.BioSequence,
 
 		// log.Println(sequence.Id(),cw[j], maxe)
 		if runExact || (atMost <= (maxe + 1)) {
+			// if true {
 			lcs, alilength := obialign.FastLCSScore(sequence, ref, maxe+1, &matrix)
+			// fmt.Println(j, cw[j], lcs, alilength, alilength-lcs)
 			// lcs, alilength := obialign.LCSScore(sequence, ref, maxe+1, matrix)
 			n++
 			if lcs == -1 {