This file is indexed.

/usr/bin/sga-astat is in sga 0.10.15-3.

This file is owned by root:root, with mode 0o755.

The actual contents of the file can be viewed below.

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
#! /usr/bin/env python
#
# sga-astat.py - Compute Myers' a-statistic for a set of contigs using read alignments
# in a bam file
#
import pysam
import sys
import getopt
import math

class ContigData:
    def __init__(self, name, len):
        self.name = name
        self.len = len
        self.nlen = len
        self.n = 0
        self.astat = 0.0
        self.bUnique = True

def computeAStat(arrivalRate, len, n):
    return arrivalRate * len - (n * math.log(2))

# Params
numContigsForInitialEstimate = 20
numIterations = 3
singleCopyThreshold = 30
bKeepDuplicates = True
minLength = 0
genomeSize = 0
arrivalRate = 0

def usage():
    print 'usage: sga-astat.py in.bam'
    print 'Compute Myers\' a-statistic for a set of contigs using the read alignments in in.bam'
    print 'Options:'
    print '    -m=INT          only compute a-stat for contigs at least INT bases in length'
    print '    -b=INT          use the longest INT contigs to perform the initial estimate'
    print '                    of the arrival rate (default: ' + str(numContigsForInitialEstimate) + ')' 
    print '    -n=INT          perform INT bootstrap iterations of the estimate'
    print '    -g=INT          use INT as the genome size instead of estimating it'
    print '    --no-duplicates do not use duplicate reads to calculate statistics'

try:
    opts, args = getopt.gnu_getopt(sys.argv[1:], 'm:b:n:g:', ['help', 'no-duplicates'])
except getopt.GetoptError, err:
        print str(err)
        usage()
        sys.exit(2)

for (oflag, oarg) in opts:
        if oflag == '-m':
            minLength = int(oarg)
        if oflag == '-b':
            numContigsForInitialEstimate = int(oarg)
        if oflag == '-n':
            numIterations = int(oarg)
        if oflag == '-g':
            genomeSize = int(oarg)
        if oflag == '--no-duplicates':
            bKeepDuplicates = False
        if oflag == '--help':
            usage()
            sys.exit(1)

if len(args) == 0:
    print 'Error: a BAM file must be provided\n'
    usage()
    sys.exit(2)

bamFilename = args[0]

sys.stderr.write('Reading alignments from ' + bamFilename + '\n')
contigData = list()

# Read the contig names and lengths from the bam
bamFile = pysam.Samfile(bamFilename, "rb")

for (name, length) in zip(bamFile.references, bamFile.lengths):
    t = name, length, 0
    contigData.append(ContigData(name, length))
    #print 'Name: ' + name
    #print 'Length: ' + str(len)

# Read the alignments and populate the counts
totalReads = 0
sumReadLength = 0

last_ref_idx = -1
last_pos = -1

for alignment in bamFile:

    if alignment.is_unmapped:
        continue

    ref_idx = alignment.rname
    ref_name = bamFile.getrname(ref_idx)

    pos = alignment.pos

    if ref_idx == last_ref_idx and pos == last_pos and not bKeepDuplicates:
        continue

    # Update the count
    cd = contigData[ref_idx]
    cd.n += 1
    totalReads += 1
    sumReadLength += alignment.rlen
    assert(cd.name == ref_name)
    last_ref_idx = ref_idx
    last_pos = pos

avgReadLen = sumReadLength / totalReads
contigData.sort(key=lambda c : c.len, reverse=True)

# Compute the length of the contigs in number of positions
# that can generate a read of length avgReadLen. Using
# when calculating the expected number of reads is a better
# approximation for small contigs
for cd in contigData:
    cd.nlen = cd.len - avgReadLen + 1 if (cd.len > avgReadLen) else 0

# Estimate the initial arrival rate using the longest contigs
# if the genome size was not provided
bootstrapLen = 0;
bootstrapReads = 0;

if genomeSize == 0:
    for i in range(0, min(numContigsForInitialEstimate, len(contigData))):
        cd = contigData[i]
        bootstrapLen += cd.nlen
        bootstrapReads += cd.n

    arrivalRate = float(bootstrapReads) / float(bootstrapLen)
    genomeSize = int(totalReads / arrivalRate)

    sys.stderr.write('Initial arrival rate: ' + str(arrivalRate) + '\n');
    sys.stderr.write('Initial genome size estimate: ' + str(genomeSize) + '\n')

    for i in range(0, numIterations):

        bootstrapLen = 0;
        bootstrapReads = 0;
        for cd in contigData:
            cd.astat = computeAStat(arrivalRate, cd.nlen, cd.n)
            cd.bUnique = cd.astat > singleCopyThreshold

            if cd.len >= minLength and cd.bUnique:
                bootstrapLen += cd.nlen
                bootstrapReads += cd.n

        # Estimate arrival rate based on unique contigs
        arrivalRate = float(bootstrapReads) / float(bootstrapLen)
        genomeSize = int(totalReads / arrivalRate)
        sys.stderr.write('Iteration ' + str(i) + ' arrival rate: ' + str(arrivalRate) + '\n');
        sys.stderr.write('Iteration ' + str(i) + ' genome size estimate: ' + str(genomeSize) + '\n')

arrivalRate = float(totalReads) / genomeSize
sys.stderr.write('Using genome size: ' + str(genomeSize) + '\n')
sys.stderr.write('Using arrival rate: ' + str(arrivalRate) + '\n')

# Compute final astat values
for cd in contigData:
    cd.astat = computeAStat(arrivalRate, cd.nlen, cd.n)
    cd.bUnique = cd.astat > singleCopyThreshold

    if cd.len >= minLength and cd.bUnique:
        bootstrapLen += cd.nlen
        bootstrapReads += cd.n

sumUnique = 0
sumRepeat = 0
for cd in contigData:
    if cd.len >= minLength and cd.nlen > 0:
        print '%s\t%d\t%d\t%d\t%f\t%f' % (cd.name, cd.len, cd.nlen, cd.n, cd.n / (cd.nlen * arrivalRate), cd.astat)
        
        if cd.bUnique:
            sumUnique += cd.len
        else:
            sumRepeat += cd.len

sys.stderr.write('Sum unique bases in contigs >= %d bp in length: %d\n' % (minLength, sumUnique))
sys.stderr.write('Sum repeat bases in contigs >= %d bp in length: %d\n' % (minLength, sumRepeat))