Merge pull request #243 from dcolligan/bam2bam

Add datadriven BAM files

scripts/bam2bam.py

@@ -0,0 +1,21 @@
+"""
+Convert a BAM file to a small BAM file
+"""
+from __future__ import division
+from __future__ import print_function
+from __future__ import unicode_literals
+
+import utils
+
+
+@utils.Timed()
+def main():
+    tool = utils.AlignmentFileTool(
+        utils.AlignmentFileConstants.BAM,
+        utils.AlignmentFileConstants.BAM)
+    tool.parseArgs()
+    tool.convert()
+
+
+if __name__ == '__main__':
+    main()

scripts/bam2sam.py

@@ -5,42 +5,16 @@
 from __future__ import print_function
 from __future__ import unicode_literals
 
-import argparse
-
-import pysam
-
 import utils
 
 
-def parseArgs():
-    parser = argparse.ArgumentParser(
-        description="BAM to SAM conversion tool")
-    parser.add_argument(
-        "inputFile", help="the name of the BAM file to read")
-    parser.add_argument(
-        "--outputFile", "-o", default='out.sam',
-        help="the name of the SAM file to write")
-    parser.add_argument(
-        "--numLines", "-n", default=10,
-        help="the number of lines to write")
-    args = parser.parse_args()
-    return args
-
-
-def bam2sam(args):
-    bam = pysam.AlignmentFile(args.inputFile, "rb")
-    sam = pysam.AlignmentFile(args.outputFile, "wh", header=bam.header)
-    for _ in xrange(args.numLines):
-        alignedSegment = bam.next()
-        sam.write(alignedSegment)
-    bam.close()
-    sam.close()
-
-
 @utils.Timed()
 def main():
-    args = parseArgs()
-    bam2sam(args)
+    tool = utils.AlignmentFileTool(
+        utils.AlignmentFileConstants.BAM,
+        utils.AlignmentFileConstants.SAM)
+    tool.parseArgs()
+    tool.convert()
 
 
 if __name__ == '__main__':

scripts/utils.py

@@ -5,6 +5,7 @@
 from __future__ import print_function
 from __future__ import unicode_literals
 
+import argparse
 import functools
 import os
 import shlex
@@ -15,6 +16,7 @@
 import humanize
 import requests
 import yaml
+import pysam
 
 
 def log(message):
@@ -130,3 +132,79 @@ def getYamlDocument(filePath):
     with open(filePath) as stream:
         doc = yaml.load(stream)
         return doc
+
+
+class AlignmentFileConstants(object):
+    """
+    A container class for constants dealing with alignment files
+    """
+    SAM = "SAM"
+    BAM = "BAM"
+    BAI = "BAI"
+
+
+class AlignmentFileTool(object):
+    """
+    Helps with operations on BAM and SAM files
+    """
+    def __init__(self, inputFileFormat, outputFileFormat):
+        self.inputFileFormat = inputFileFormat
+        self.outputFileFormat = outputFileFormat
+        self.args = None
+
+    def parseArgs(self):
+        description = "{} to {} conversion tool".format(
+            self.inputFileFormat, self.outputFileFormat)
+        parser = argparse.ArgumentParser(
+            description=description)
+        inputHelpText = "the name of the {} file to read".format(
+            self.inputFileFormat)
+        parser.add_argument(
+            "inputFile", help=inputHelpText)
+        outputHelpText = "the name of the {} file to write".format(
+            self.outputFileFormat)
+        defaultOutputFilePath = "out.{}".format(
+            self.outputFileFormat.lower())
+        parser.add_argument(
+            "--outputFile", "-o", default=defaultOutputFilePath,
+            help=outputHelpText)
+        parser.add_argument(
+            "--numLines", "-n", default=10,
+            help="the number of lines to write")
+        parser.add_argument(
+            "--skipIndexing", default=False, action='store_true',
+            help="don't create an index file")
+        args = parser.parse_args()
+        self.args = args
+
+    def convert(self):
+        # set flags
+        if self.inputFileFormat == AlignmentFileConstants.SAM:
+            inputFlags = "r"
+        elif self.inputFileFormat == AlignmentFileConstants.BAM:
+            inputFlags = "rb"
+        if self.outputFileFormat == AlignmentFileConstants.SAM:
+            outputFlags = "wh"
+        elif self.outputFileFormat == AlignmentFileConstants.BAM:
+            outputFlags = "wb"
+        # open files
+        inputFile = pysam.AlignmentFile(
+            self.args.inputFile, inputFlags)
+        outputFile = pysam.AlignmentFile(
+            self.args.outputFile, outputFlags, header=inputFile.header)
+        outputFilePath = outputFile.filename
+        log("Creating alignment file '{}'".format(outputFilePath))
+        # write new file
+        for _ in xrange(self.args.numLines):
+            alignedSegment = inputFile.next()
+            outputFile.write(alignedSegment)
+        # clean up
+        inputFile.close()
+        outputFile.close()
+        # create index file
+        if (not self.args.skipIndexing and
+                self.outputFileFormat == AlignmentFileConstants.BAM):
+            indexFilePath = "{}.{}".format(
+                outputFilePath, AlignmentFileConstants.BAI.lower())
+            log("Creating index file '{}'".format(indexFilePath))
+            runCommand("samtools index {}".format(outputFilePath))

tests/data/reads/README.txt

@@ -0,0 +1,5 @@
+Files in the directories wgBam and wgSam are taken from here:
+
+http://hgdownload.cse.ucsc.edu/goldenPath/hg19/encodeDCC/wgEncodeUwRepliSeq/
+
+And shortened / converted with tools in the scripts/ directory.

tests/datadriven/test_reads.py

@@ -26,4 +26,5 @@ def getProtocolClass(self):
 
     def testFixMe(self):
         readGroupSet = self._gaObject
+        # TODO
         self.assertIsNotNone(readGroupSet)