# Adding entries to bigwig file

I generate bigwig files using a shell script based on bedtools genomecov (to generate a bedgraph from a bam file) bedmap (to compute means across 10 bp bins) and bedGraphToBigWig to convert the binned bedgraph into bigwig.

Sometimes, the bam file has no data on some chromosomes, this results in a bedgraph file with no entries for the empty chromosomes.

I would like to add the missing entries. How could I do this?

I tried to see what I could do with pyBigWig, opening the bedgraph file in "a" mode, but don't really know how to proceed:

# I have the chromosome sizes for my genome of interest already in a list:
chrom_sizes
#[('I', 15072434),
# ('II', 15279421),
# ('III', 13783801),
# ('IV', 17493829),
# ('MtDNA', 13794),
# ('V', 20924180),
# ('X', 17718942)]
bw_test = pyBigWig.open("path/to/bigwig", "a")
bw_test.chroms()
# {'IV': 17493829, 'V': 20924180, 'X': 17718942}
bw_test.addHeader(chrom_sizes)
# RuntimeError: Received an error in bwCreateHdr


## 1 Answer

Presuming you still have your BAM file sitting around somewhere:

#!/usr/bin/env python
import pysam
import pyBigWig

bam = pysam.AlignmentFile("alignments.bam")
ifile = pyBigWig.open("your current file.bw")
ofile = pyBigWig.open("your new file.bw", "w")

# Create/Add the header
hdr = []
for chrom in bam.header['SQ']:
hdr.append((chrom['SN'], chrom['LN']))
bam.close()
ofile.addHeader(hdr)

# Write the bigWig
for _ in hdr:
try:
ints = ifile.intervals(_[0])
starts = [x[0] for x in ints]
ends = [x[1] for x in ints]
vals = [x[2] for x in ints]
ofile.addEntries([_[0]] * len(starts), starts, ends=ends, values=vals)
except:
# intervals() will throw an error if the chromosome isn't present
pass
ofile.close()
ifile.close()


The reason you can't append to a bigWig file is that it has a various indices and zoom levels typically built in. You'd have to add to them without overwriting neighboring blocks of data. In theory this might sometimes be theoretically possible, but not typically and I'm not aware of tools that support this.

• If I understand correctly, the bam file is not necessary in my case, because I already have the chrom_sizes list, which should be the same as the hdr you obtain by parsing the bam header.
– bli
Oct 31 '17 at 8:38
• Correct, if you already have such a list then you don’t need a BAM file. Oct 31 '17 at 8:42
• Just a detail: It seems strange to me to use a _ placeholder as looping variable if you then use it to access its first element. I suppose for (chrom, _) in hdr: would work as well while avoiding you the _[0].
– bli
Oct 31 '17 at 8:44
• Yup that’d work too. Oct 31 '17 at 8:48