support gzip format

.gitignore

@@ -9,3 +9,6 @@ build/
 
 # Directory created by dartdoc
 doc/api/
+
+# Benchmark test
+benchmark/

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 0.0.4
+
+- support gz compression format
+
 ## 0.0.3
 
 - filter sequence by motif (`--filter-motif`)

Makefile

@@ -0,0 +1,14 @@
+#
+# Makefile
+# Ye Chang, 2020-04-07 21:23
+#
+
+all: binary
+	@echo "Done!"
+
+binary:
+	@echo "Building binary by dart2native..."
+	@dart2native bin/main.dart -o build/bio 1>/dev/null 2>/dev/null
+
+# vim:ft=make
+#

README.md

@@ -66,5 +66,9 @@ Functions and command line tools for biological computation written in **Dart**.
 
 ## BENCHMARK
 
-vs. biopython
-vs. seqkit
+| tool            | test IO time | test RC time |
+| --------------- | -----------: | -----------: |
+| biodart         |        6.678 |       15.533 |
+| seqkit (Golang) |        0.996 |        0.879 |
+| seqtk (C)       |        0.849 |        0.854 |
+

TODO.md

@@ -1,11 +1,14 @@
 > seqIO
 
-- [x] support stdin and stdout
+- [ ] support gzip format as output
+- [ ] improve speed
 - [ ] split record name and record description
 - [ ] support multiple line fastq?
 - [ ] trim low quality base
 - [ ] population genetic calculator
 - [ ] support degenerate bases matching
+- [x] support stdin and stdout
+- [x] support gzip format as input
 
 > alignIO
 

bin/main.dart

@@ -40,6 +40,8 @@ class SeqCommand extends Command {
         outputFile: argResults['output'],
         inputFormat: argResults['input-format'],
         outputFormat: argResults['output-format'],
+        inputCompressed: argResults['input-compressed'],
+        outputCompressed: argResults['output-compressed'],
         fastaLineLength: int.parse(argResults['fasta-line-length']),
         sample: int.parse(argResults['sample']),
         randomSeed: int.tryParse(argResults['sample-seed']),
@@ -60,8 +62,12 @@ class SeqCommand extends Command {
           abbr: 'o', defaultsTo: '-', valueHelp: 'Path of output file')
       ..addOption('input-format',
           abbr: 's', help: 'Format of input file', valueHelp: 'auto')
+      ..addOption('input-compressed',
+          help: 'Whether input file is in .gz format', valueHelp: 'auto')
       ..addOption('output-format',
           abbr: 't', help: 'Format of output file', valueHelp: 'auto')
+      ..addOption('output-compressed',
+          help: 'Whether output file is in .gz format', valueHelp: 'auto')
       ..addOption('fasta-line-length',
           abbr: 'l', defaultsTo: '0', help: 'Number of charaters in each line')
       ..addOption('sample',

lib/seq.dart

@@ -1,4 +1,4 @@
-import 'dart:io' show File, exit;
+import 'dart:io' show File, exit, gzip;
 import 'dart:convert' show LineSplitter, utf8;
 import 'dart:math' show Random, min;
 import 'package:bio/utils.dart' as utils;
@@ -117,9 +117,18 @@ Stream<Seq> _fqStringStream2SeqStream(Stream<String> lines) async* {
 }
 
 /// Read file as Stream<Seq>
-Stream<Seq> file2Stream(File file, String format) {
+Stream<Seq> file2Stream(File file, String format, {bool isGzip = false}) {
   // Read file as Stream<String>
-  var lines = file.openRead().transform(utf8.decoder).transform(LineSplitter());
+  Stream<String> lines;
+  if (isGzip) {
+    lines = file
+        .openRead()
+        .transform(gzip.decoder)
+        .transform(utf8.decoder)
+        .transform(LineSplitter());
+  } else {
+    lines = file.openRead().transform(utf8.decoder).transform(LineSplitter());
+  }
   // parse Stream<String> into Stream<Seq>
   Stream<Seq> s;
   if (format == 'fa') {
@@ -175,8 +184,10 @@ int countFq(File file) {
 void seqIO(
     {String inputFile,
     String outputFile,
+    bool inputCompressed,
     String inputFormat,
     String outputFormat,
+    bool outputCompressed,
     int fastaLineLength,
     int sample = 0,
     int randomSeed,
@@ -215,13 +226,23 @@ void seqIO(
     'fastq': 'fq',
     'fq': 'fq'
   };
-  inputFormat ??= inputFile.split('.').last;
+
+  bool isGzip;
+  if (inputFile.split('.').last == 'gz') {
+    isGzip = true;
+    var nameSplited = inputFile.split('.');
+    inputFormat ??= nameSplited[nameSplited.length - 2];
+  } else {
+    isGzip = false;
+    inputFormat ??= inputFile.split('.').last;
+  }
   if (supportedFormats.containsKey(inputFormat)) {
     inputFormat = supportedFormats[inputFormat];
   } else {
     log.warning('${inputFormat} format is not supported!');
     exit(1);
   }
+
   outputFormat ??= outputFile.split('.').last;
   if (supportedFormats.containsKey(outputFormat)) {
     outputFormat = supportedFormats[outputFormat];
@@ -231,7 +252,7 @@ void seqIO(
   }
 
   // Read file as Stream<Seq>
-  var inputStream = file2Stream(infile, inputFormat);
+  var inputStream = file2Stream(infile, inputFormat, isGzip: isGzip);
 
   // var castStream = inputStream.asBroadcastStream();
   // castStream.length.then((value) => print("stream.length: $value"));

pubspec.yaml

@@ -1,6 +1,6 @@
 name: bio
 description: "Some useful command line utils written in Dart. For parsing and manipulating sequence records in various formats..."
-version: 0.0.3
+version: 0.0.4
 homepage: https://github.com/yech1990/biodart
 documentation: https://pub.dev/documentation/bio