Bam to bigwig

[gokhulkrishnakilaru]

Hi,

I have a bunch of sorted bam files and would like to get them converted into bigwig files. Any ideas are highly appreciated.

Thanks in advance.

[maubp]

I use this: http://github.com/chapmanb/bcbb/blob...m_to_wiggle.py

See also http://biostar.stackexchange.com/que...-sam-to-wiggle and Google.

[Richard Finney]

What's needed is "samtools depth" combined with "UCSC wigToBigWig" (that doesn't use 32 GB of memory.)

[GW_OK]

I use a combo of samtools, bedtools, and UCSC scripts with a few steps in between to remove chimeras:
samtools sort -n file.bam file_name_sorted
samtools view -uf 0x2 ./file_name_sorted.bam | bamToBed -i stdin -bedpe > file.bedpe
awk '$1 == $4' file.bedpe | awk '{OFS="\t"; print $1, $2, $6, $7}' | sort -k 1,1 > file.bed
genomeCoverageBed -i file.bed -g hg19.genome -bg > file_name_sorted.cov
bedGraphToBigWig file_name_sorted.cov hg19.genome file.bw

[gokhulkrishnakilaru]

Thank you all.

Will give a try and shall let u know.

[aleferna]

Another alternative

For RNA seq data (which has intron spanning reads) I made the following script which:
1. Does not report coverage over introns.
2. Generates one file for each strand
3. Summarized data in 10 bp bins
4. Does not report bins with less than 3 reads.


here $bam is the full bam file and $base is the shorted sample name.

samtools mpileup -d 100000 -q 10 --rf "REVERSE" --ff "UNMAP,SECONDARY,QCFAIL,DUP" $bam | awk -F '\t' '{
curbin = "chr"$1"\t"int($2/10)*10"\t"int($2/10)*10+10
if (curbin != lastbin ){
if (tot/cts > 3){
print lastbin"\t"tot/cts*-1;
}
cts=0;
tot=0;
lastbin = curbin
}
a=$5
gsub(/<|>/,"",a)
a=$4 - length($5) + length(a)
cts = cts + 1;
tot = tot + a
}' > washu/bg/$base.Rev.bedGraph &

samtools mpileup -d 100000 -q 10 --ff "REVERSE,UNMAP,SECONDARY,QCFAIL,DUP" $bam | awk -F '\t' '{
curbin = "chr"$1"\t"int($2/10)*10"\t"int($2/10)*10+10
if (curbin != lastbin ){
if (tot/cts > 3){
print lastbin"\t"tot/cts;
}
cts=0;
tot=0;
lastbin = curbin
}
a=$5
gsub(/<|>/,"",a)
a=$4 - length($5) + length(a)
cts = cts + 1;
tot = tot + a
}' > washu/bg/$base.For.bedGraph &

[dpryan]

You could also just use bamCoverage from deepTools, which handles spliced reads as well.