stats: add option to output in machine-friendly tabular format

shenwei356 · Sep 16, 2017 · a6aae0f · a6aae0f
1 parent 32a6d7a
commit a6aae0f
Show file tree

Hide file tree

Showing 3 changed files with 74 additions and 4 deletions.
diff --git a/dev-version.md b/dev-version.md
@@ -1,3 +1,5 @@
 - `seqkit convert`: fix bug of read quality containing only 3 or less values.
   [shenwei356/bio/issues/3](https://github.com/shenwei356/bio/issues/3)
+- `seqkit stats`: add option `-T/--tabular` to output in machine-friendly tabular format
+  [#23](https://github.com/shenwei356/seqkit/issues/23)
 - fix typo [#22](https://github.com/shenwei356/seqkit/issues/22)
diff --git a/doc/docs/usage.md b/doc/docs/usage.md
@@ -573,18 +573,19 @@ Examples
 Usage
 
 ```
-simple statistics of FASTA files
+simple statistics of FASTA/Q files
 
 Usage:
   seqkit stats [flags]
 
 Aliases:
   stats, stat
 
-
 Flags:
-  -a, --all                  all statistics, including sum_gap, N50, L50
+  -a, --all                  all statistics, including sum_gap, N50
   -G, --gap-letters string   gap letters (default "- .")
+  -h, --help                 help for stats
+  -T, --tabular              output in machine-friendly tabular format
 
 ```
 
@@ -599,6 +600,25 @@ Eexamples
         reads_1.fq.gz  FASTQ   DNA      2,500    567,516      226      227      229
         reads_2.fq.gz  FASTQ   DNA      2,500    560,002      223      224      225
 
+1. Machine-friendly tabular format
+
+        $ seqkit stats *.f{a,q}.gz -T
+        file    format  type    num_seqs        sum_len min_len avg_len max_len
+        hairpin.fa.gz   FASTA   RNA     28645   2949871 39      103.0   2354
+        mature.fa.gz    FASTA   RNA     35828   781222  15      21.8    34
+        Illimina1.8.fq.gz       FASTQ   DNA     10000   1500000 150     150.0   150
+        reads_1.fq.gz   FASTQ   DNA     2500    567516  226     227.0   229
+        reads_2.fq.gz   FASTQ   DNA     2500    560002  223     224.0   225
+
+        $ seqkit stats *.f{a,q}.gz -T | csvtk pretty -t
+        file                format   type   num_seqs   sum_len   min_len   avg_len   max_len
+        hairpin.fa.gz       FASTA    RNA    28645      2949871   39        103.0     2354
+        mature.fa.gz        FASTA    RNA    35828      781222    15        21.8      34
+        Illimina1.8.fq.gz   FASTQ    DNA    10000      1500000   150       150.0     150
+        reads_1.fq.gz       FASTQ    DNA    2500       567516    226       227.0     229
+        reads_2.fq.gz       FASTQ    DNA    2500       560002    223       224.0     225
+
+
 1. Extra information
 
         $ seqkit stats *.f{a,q}.gz -a

diff --git a/seqkit/cmd/stat.go b/seqkit/cmd/stat.go
@@ -25,6 +25,7 @@ import (
 	"io"
 	"runtime"
 	"sort"
+	"strings"
 
 	"github.com/cznic/sortutil"
 	"github.com/dustin/go-humanize"
@@ -66,6 +67,7 @@ var statCmd = &cobra.Command{
 		gapLettersBytes := []byte(gapLetters)
 
 		all := getFlagBool(cmd, "all")
+		tabular := getFlagBool(cmd, "tabular")
 
 		files := getFileList(args)
 
@@ -154,6 +156,51 @@ var statCmd = &cobra.Command{
 			}
 		}
 
+		// tabular output
+		if tabular {
+			colnames := []string{
+				"file",
+				"format",
+				"type",
+				"num_seqs",
+				"sum_len",
+				"min_len",
+				"avg_len",
+				"max_len",
+			}
+			if all {
+				colnames = append(colnames, []string{"sum_gap", "N50"}...)
+			}
+			outfh.WriteString(strings.Join(colnames, "\t") + "\n")
+
+			for _, info := range statInfos {
+				if !all {
+					outfh.WriteString(fmt.Sprintf("%s\t%s\t%s\t%d\t%d\t%d\t%.1f\t%d\n",
+						info.file,
+						info.format,
+						info.t,
+						info.num,
+						info.lenSum,
+						info.lenMin,
+						info.lenAvg,
+						info.lenMax))
+				} else {
+					outfh.WriteString(fmt.Sprintf("%s\t%s\t%s\t%d\t%d\t%d\t%.1f\t%d\t%d\t%d\n",
+						info.file,
+						info.format,
+						info.t,
+						info.num,
+						info.lenSum,
+						info.lenMin,
+						info.lenAvg,
+						info.lenMax,
+						info.gapSum,
+						info.N50))
+				}
+			}
+			return
+		}
+
 		// format output
 		columns := []prettytable.Column{
 			{Header: "file"},
@@ -226,6 +273,7 @@ type statInfo struct {
 func init() {
 	RootCmd.AddCommand(statCmd)
 
+	statCmd.Flags().BoolP("tabular", "T", false, "output in machine-friendly tabular format")
 	statCmd.Flags().StringP("gap-letters", "G", "- .", "gap letters")
-	statCmd.Flags().BoolP("all", "a", false, "all statistics, including sum_gap, N50, L50")
+	statCmd.Flags().BoolP("all", "a", false, "all statistics, including sum_gap, N50")
 }