add kmer match and query ratio search options

api/server.go

@@ -145,6 +145,8 @@ func searchFastq(w http.ResponseWriter, r *http.Request) {
 		OutFormat:        "tsv",
 		MaxResults:       10,
 		ExtractPositions: false,
+		MinKMatch:        10,
+		MinKRatio:        0.05,
 		SubMatrix:        "blosum62",
 		GapOpen:          11,
 		GapExtend:        1,
@@ -170,6 +172,8 @@ func searchNucleotide(w http.ResponseWriter, r *http.Request) {
 		OutFormat:        "tsv",
 		MaxResults:       10,
 		ExtractPositions: false,
+		MinKMatch:        10,
+		MinKRatio:        0.05,
 		SubMatrix:        "blosum62",
 		GapOpen:          11,
 		GapExtend:        1,
@@ -196,6 +200,8 @@ func searchProtein(w http.ResponseWriter, r *http.Request) {
 		OutFormat:        "tsv",
 		MaxResults:       10,
 		ExtractPositions: false,
+		MinKMatch:        10,
+		MinKRatio:        0.05,
 		SubMatrix:        "blosum62",
 		GapOpen:          11,
 		GapExtend:        1,
@@ -278,6 +284,18 @@ func parseSearchOptions(searchOpts *search.SearchOptions, w http.ResponseWriter,
 		searchOpts.Align = true
 	}
 
+	if r.FormValue("minkmatch") != "10" {
+		if minKMatch, err := strconv.ParseInt(r.FormValue("minkmatch"), 10, 64); err == nil {
+			searchOpts.MinKMatch = minKMatch
+		}
+	}
+
+	if r.FormValue("minkratio") != "0.05" {
+		if minKRatio, err := strconv.ParseFloat(r.FormValue("minkratio"), 64); err == nil {
+			searchOpts.MinKRatio = minKRatio
+		}
+	}
+
 	if strings.ToLower(r.FormValue("sub-matrix")) != "blosum62" {
 		searchOpts.SubMatrix = strings.ToLower(r.FormValue("sub-matrix"))
 	}

cmd/kaamer/main.go

@@ -71,6 +71,9 @@ func main() {
 
     // aln options
 
+      -mink         minimum number of k-mer matches to report a hit (default: 10)
+      -minr         minimum ratio of query k-mer matches to report a hit (default: 0.05)
+
       -mat          substitution matrix (default: BLOSUM62)
       -gop          gap open penalty (default: 11)
       -gex          gap extension penalty (default: 1)
@@ -90,6 +93,8 @@ func main() {
 	var addAnnotation = flag.Bool("ann", false, "add annotation flag")
 	var addPositions = flag.Bool("pos", false, "add position flag")
 
+	var minKMatch = flag.Int64("mink", 10, "minimum number of k-mer matches to report a hit")
+	var minKRatio = flag.Float64("minr", 0.05, "minimum ratio of query k-mer matches to report a hit")
 	var subMatrix = flag.String("mat", "blosum62", "substitution matrix")
 	var gapOpen = flag.Int("gop", 11, "gap open penalty")
 	var gapExtend = flag.Int("gex", 1, "gap extension penalty")
@@ -163,6 +168,8 @@ func main() {
 		options.Align = *addAlignment
 		options.ExtractPositions = *addPositions
 		options.Annotations = *addAnnotation
+		options.MinKMatch = *minKMatch
+		options.MinKRatio = *minKRatio
 		options.SubMatrix = *subMatrix
 		options.GapOpen = *gapOpen
 		options.GapExtend = *gapExtend

docs/client.md

@@ -42,6 +42,9 @@ The kaamer CLI is a client to query (-search) a kaamer database.
 
    // aln options
 
+      -mink         minimum number of k-mers match to report a hit (default: 10)
+      -minr         minimum ratio of query k-mers matches to report a hit (default: 0.05)
+
       -mat          substitution matrix (default: BLOSUM62)
       -gop          gap open penalty (default: 11)
       -gex          gap extension penalty (default: 1)
@@ -67,7 +70,7 @@ The kaamer CLI is a client to query (-search) a kaamer database.
 * -g Genetic Code
 
    Genetic code number for translated search (with -t fastq or -t nt) \
-   One of the following : 1-15 except 7,8 (default 11 - bacteria) \
+   One of the following : 1-15 except 7,8 (default: 11 - bacteria) \
    See https://www.bioinformatics.org/JaMBW/2/3/TranslationTables.html
 
 * -i Input File
@@ -76,7 +79,7 @@ The kaamer CLI is a client to query (-search) a kaamer database.
 
 * -m Max Results
 
-    Maximum number of results to return (default 10) 
+    Maximum number of results to return (default: 10) 
 
 * -o Outpout
 
@@ -92,18 +95,19 @@ The kaamer CLI is a client to query (-search) a kaamer database.
 
 * -ann Hit Annotations
 
-    Add hit annotations output (default false)
+    Add hit annotations output (default: false)
 
 * -pos Positions Match
 
-    Add the positions that has a match with the hit (default false) 
+    Add the positions that has a match with the hit (default: false) 
 
 ##### Alignment Options
 
+* -mink  Minimum number of k-mers match to report a hit (default: 10)
+* -minr  Minimum ratio of query k-mers matches to report a hit (default: 0.05)
+
 * -mat  Substitution matrix (default: BLOSUM62)
-
 * -gop  Gap open penalty (default: 11)
-
 * -gex  Gap extension penalty (default: 1)
 
 

pkg/search/search.go

@@ -48,9 +48,9 @@ const (
 )
 
 var (
-	dbStats     kvstore.KStats
-	kMatchRatio = 0.05      // at least 5% of kmer hits (on query)
-	minKMatch   = int64(10) // at least 10 kmer hits
+	dbStats kvstore.KStats
+	// kMatchRatio = 0.05      // at least 5% of kmer hits (on query)
+	// minKMatch   = int64(10) // at least 10 kmer hits
 )
 
 type SearchOptions struct {
@@ -66,6 +66,8 @@ type SearchOptions struct {
 	SubMatrix        string
 	GapOpen          int
 	GapExtend        int
+	MinKMatch        int64
+	MinKRatio        float64
 }
 
 type SearchResults struct {
@@ -186,11 +188,13 @@ func NewSearchResult(searchOptions SearchOptions, _dbStats kvstore.KStats, kvSto
 
 func (queryResult *QueryResult) FilterResults(searchOptions SearchOptions) {
 
+	fmt.Printf("KmerMatch: %d  |  KmerRatio: %f \n", searchOptions.MinKMatch, searchOptions.MinKRatio)
+
 	var hitsToDelete []uint32
 	var lastGoodHitPosition = len(queryResult.SearchResults.Hits) - 1
 
 	for i, hit := range queryResult.SearchResults.Hits {
-		if (float64(hit.Kmatch)/float64(queryResult.Query.SizeInKmer)) < kMatchRatio || hit.Kmatch < minKMatch {
+		if (float64(hit.Kmatch)/float64(queryResult.Query.SizeInKmer)) < searchOptions.MinKRatio || hit.Kmatch < searchOptions.MinKMatch {
 			if lastGoodHitPosition == (len(queryResult.SearchResults.Hits) - 1) {
 				lastGoodHitPosition = i - 1
 			}

pkg/search/search_fastq.go

@@ -116,7 +116,7 @@ func FastqSearch(searchOptions SearchOptions, kvStores *kvstore.KVStores, nbOfTh
 					wgMP.Wait()
 
 					searchRes.Hits = sortMapByValue(searchRes.Counter.GetCountersMap())
-					if len(searchRes.Hits) > 0 && searchRes.Hits[0].Kmatch >= minKMatch {
+					if len(searchRes.Hits) > 0 && searchRes.Hits[0].Kmatch >= searchOptions.MinKMatch {
 						qR = QueryResult{Query: q, SearchResults: searchRes, HitEntries: map[uint32]kvstore.Protein{}}
 						SetBestStartCodon(&qR)
 						qR.FilterResults(searchOptions)

pkg/search/search_nucleotide.go

@@ -113,7 +113,7 @@ func NucleotideSearch(searchOptions SearchOptions, kvStores *kvstore.KVStores, n
 					wgMP.Wait()
 
 					searchRes.Hits = sortMapByValue(searchRes.Counter.GetCountersMap())
-					if len(searchRes.Hits) > 0 && searchRes.Hits[0].Kmatch >= minKMatch {
+					if len(searchRes.Hits) > 0 && searchRes.Hits[0].Kmatch >= searchOptions.MinKMatch {
 						qR := QueryResult{Query: q, SearchResults: searchRes, HitEntries: map[uint32]kvstore.Protein{}}
 						SetBestStartCodon(&qR)
 						qR.FilterResults(searchOptions)

pkg/searchcli/searchcli.go

@@ -50,6 +50,8 @@ func NewSearchRequest(options SearchRequestOptions) {
 	bodyWriter.WriteField("align", strconv.FormatBool(options.Align))
 	bodyWriter.WriteField("annotations", strconv.FormatBool(options.Annotations))
 	bodyWriter.WriteField("positions", strconv.FormatBool(options.ExtractPositions))
+	bodyWriter.WriteField("minkmatch", strconv.FormatInt(options.MinKMatch, 10))
+	bodyWriter.WriteField("minkratio", fmt.Sprintf("%f", options.MinKRatio))
 	bodyWriter.WriteField("sub-matrix", options.SubMatrix)
 	bodyWriter.WriteField("gap-open", strconv.Itoa(options.GapOpen))
 	bodyWriter.WriteField("gap-extend", strconv.Itoa(options.GapExtend))