@@ -42,9 +42,6 @@ var commonCmd = &cobra.Command{
4242
4343` ,
4444 Run : func (cmd * cobra.Command , args []string ) {
45- if len (args ) < 2 {
46- checkError (errors .New ("at least 2 files needed" ))
47- }
4845 config := getConfigs (cmd )
4946 alphabet := config .Alphabet
5047 idRegexp := config .IDRegexp
@@ -68,21 +65,34 @@ var commonCmd = &cobra.Command{
6865 }
6966
7067 files := getFileList (args )
68+ if len (files ) < 2 {
69+ checkError (errors .New ("at least 2 files needed" ))
70+ }
7171
7272 outfh , err := xopen .Wopen (outFile )
7373 checkError (err )
7474 defer outfh .Close ()
7575
76- counter := make (map [string ]map [string ]int )
77- names := make (map [string ]map [string ]string )
76+ // target -> file -> struct{}
77+ counter := make (map [string ]map [string ]struct {})
78+ // target -> file -> seqname
79+ names := make (map [string ]map [string ][]string )
80+
7881 var fastxReader * fastx.Reader
7982
8083 // read all files
8184 var subject string
85+ var checkFirstFile = true
86+ var firstFile string
8287 for _ , file := range files {
8388 if ! quiet {
8489 log .Infof ("read file: %s" , file )
8590 }
91+ if checkFirstFile {
92+ firstFile = file
93+ checkFirstFile = false
94+ }
95+
8696 fastxReader , err = fastx .NewReader (alphabet , file , idRegexp )
8797 checkError (err )
8898 for {
@@ -124,42 +134,56 @@ var commonCmd = &cobra.Command{
124134 }
125135
126136 if _ , ok := counter [subject ]; ! ok {
127- counter [subject ] = make (map [string ]int )
137+ counter [subject ] = make (map [string ]struct {} )
128138 }
129- counter [subject ][file ] = counter [ subject ][ file ] + 1
139+ counter [subject ][file ] = struct {}{}
130140
131141 if _ , ok := names [subject ]; ! ok {
132- names [subject ] = make (map [string ]string )
142+ names [subject ] = make (map [string ][] string )
133143 }
134- names [subject ][file ] = string (record .Name )
144+ names [subject ][file ] = append ( names [ subject ][ file ], string (record .Name ) )
135145 }
136146 }
137147
138148 // find common seqs
139149 if ! quiet {
140150 log .Info ("find common seqs ..." )
141151 }
142- fileNum := len (args )
143- firstFile := args [ 0 ]
144- namesOK := make ( map [ string ] int )
145- n := 0
146- for subject , count := range counter {
147- if len (count ) != fileNum {
152+ fileNum := len (files )
153+ namesOK := make ( map [ string ] struct {})
154+ var n , n2 int
155+ var seqname string
156+ for subject , presence := range counter {
157+ if len (presence ) != fileNum {
148158 continue
149159 }
150- namesOK [ names [ subject ][ firstFile ]] = counter [ subject ][ firstFile ]
160+
151161 n ++
152- }
153- if ! quiet {
154- log .Infof ("%d common seqs found" , n )
162+ for _ , seqname = range names [subject ][firstFile ] {
163+ n2 ++
164+ namesOK [seqname ] = struct {}{}
165+ }
155166 }
156167
168+ var t string
169+ if byName {
170+ t = "sequence headers"
171+ } else if bySeq {
172+ t = "sequences"
173+ } else {
174+ t = "sequence IDs"
175+ }
157176 if n == 0 {
177+ log .Infof ("no common %s found" , t )
158178 return
159179 }
180+ if ! quiet {
181+ log .Infof ("%d unique %s found in %d files, which belong to %d records in the first file: %s" ,
182+ n , t , fileNum , len (namesOK ), firstFile )
183+ }
160184
161185 if ! quiet {
162- log .Infof ("extract common seqs from first file: %s" , firstFile )
186+ log .Infof ("extract seqs from the first file: %s" , firstFile )
163187 }
164188
165189 // extract
@@ -174,11 +198,12 @@ var commonCmd = &cobra.Command{
174198 checkError (err )
175199 break
176200 }
201+ if fastxReader .IsFastq {
202+ config .LineWidth = 0
203+ }
177204
178- name := string (record .Name )
179- if _ , ok := namesOK [name ]; ok && namesOK [name ] > 0 {
205+ if _ , ok := namesOK [string (record .Name )]; ok {
180206 record .FormatToWriter (outfh , lineWidth )
181- namesOK [name ] = 0
182207 }
183208 }
184209 },
0 commit comments