test/test_BAMBAI.py

   1 #!/usr/bin/env python
   2 # Time-stamp: <2021-03-10 18:55:41 Tao Liu>
   3
   4 import unittest
   5
   6 from MACS3.IO.BAM import *
   7
   8 class Test_BAIFile ( unittest.TestCase ):
   9
  10     def setUp ( self ):
  11         self.bamfile = "test/tiny.bam"
  12         self.baifile = "test/tiny.bam.bai"
  13         self.bai = BAIFile( self.baifile )
  14
  15     def test_load ( self ):
  16         expected = {'ref_beg': 50724864, 'ref_end': 2709258240, 'n_mapped': 971, 'n_unmapped': 0}
  17         meta = self.bai.get_metadata_by_refseq( 0 )
  18         self.assertDictEqual( expected, meta )
  19
  20     def test_get_chunks_bin ( self ):
  21         expected = [(2178362419, 2178363422)]
  22         chunks = self.bai.get_chunks_by_bin( 0, 4728 )
  23         self.assertListEqual( chunks, expected )
  24
  25     def test_get_chunks_list_bins ( self ):
  26         expected = [(50741975, 50751615), (2178362419, 2178363422), (2178363422, 2709258240)]
  27         chunks = self.bai.get_chunks_by_list_of_bins( 0, [ 591, 4694, 4728] )
  28         self.assertListEqual( chunks, expected )
  29
  30     def test_get_chunks_region ( self ):
  31         expected = [(1130775424, 1130820164), (1130820164, 1130822629), (1130822629, 2178352280), (2178352280, 2178353973), (2178353973, 2178356807), (2178356807, 2178358139), (2178358139, 2178359954), (2178359954, 2178362419), (2178362419, 2178363422), (2178363422, 2709258240)]
  32         chunks = self.bai.get_chunks_by_region( 0, 557000, 851600 )
  33         self.assertListEqual( chunks, expected )
  34
  35     def test_get_chunks_list_regions ( self ):
  36         expected = [(1130770109, 1130772126), (1130772126, 1130775424), (1130775424, 1130820164), (2178363422, 2709258240)]
  37         chunks = self.bai.get_chunks_by_list_of_regions( 0, [ (500000, 600000), (800000, 850000) ] )
  38         self.assertListEqual( chunks, expected )
  39
  40
  41 class Test_BAM_w_BAI ( unittest.TestCase ):
  42
  43     def setUp ( self ):
  44         self.bamfile = "test/tiny.bam"
  45         self.baifile = "test/tiny.bam.bai"
  46         self.bam = BAMaccessor( self.bamfile )
  47
  48     def test_get_reads_large_dedup ( self ):
  49         c = b"chr10"
  50         s = 100
  51         e = 900000
  52         a = self.bam.get_reads_in_region( c, s, e, maxDuplicate = 1 )
  53         self.assertEqual( len(a), 941 )
  54
  55     def test_get_reads_large ( self ):
  56         c = b"chr10"
  57         s = 100
  58         e = 900000
  59         a = self.bam.get_reads_in_region( c, s, e, maxDuplicate = 100 )
  60         self.assertEqual( len(a), 971 )
  61
  62     def test_get_reads_mod ( self ):
  63         c = b"chr10"
  64         s = 100000
  65         e = 550000
  66         a = self.bam.get_reads_in_region( c, s, e, maxDuplicate = 100 )
  67         self.assertEqual( len(a), 520 )
  68
  69     def test_get_reads_few( self ):
  70         c = b"chr10"
  71         s = 142999
  72         e = 163000
  73         expected = """chr10     145930  145966  MAGNUM:8:93:1052:1153#0 36      +
  74 chr10   148133  148169  MAGNUM:8:108:19082:3782#0       36      -
  75 chr10   148930  148966  MAGNUM:8:34:18198:13544#0       36      +
  76 chr10   152575  152611  ROCKFORD:1:114:13238:15292#0    36      +
  77 chr10   152927  152963  ROCKFORD:1:21:4055:6592#0       36      -
  78 chr10   153995  154031  MAGNUM:8:44:12716:12267#0       36      +
  79 chr10   156153  156189  MAGNUM:8:74:16381:18650#0       36      -
  80 chr10   159259  159295  ROCKFORD:1:112:5095:4754#0      36      +"""
  81         a = self.bam.get_reads_in_region( c, s, e, maxDuplicate = 100 )
  82         result = "\n".join( map( str, a ) )
  83         self.assertEqual( len(a), 8 )
  84         self.assertEqual( result, expected )
  85
  86     def test_get_empty1 ( self ):
  87         # region outside of bams
  88         c = b"chr10"
  89         s = 961000
  90         e = 963000
  91         a = self.bam.get_reads_in_region( c, s, e )
  92         self.assertEqual( len(a), 0 )
  93
  94     def test_get_empty2 ( self ):
  95         # small region missed in bam
  96         c = b"chr10"
  97         s = 70000
  98         e = 90000
  99         a = self.bam.get_reads_in_region( c, s, e )
 100         self.assertEqual( len(a), 0 )
 101