Increasing widths of the peaks in a bed file from ChIP-seq

Question

[morovatunc]

Hello Biostars community,

I have generated bed file from chip-seq experiment and would like to filter my mutations that occurred in these regions with bed file input. But I have found less mutations than I expected.

My initial guess is increasing the peak with by 500 bp from both sides but I have a concern before I conduct this step. I dont know if -500 from start and +500 to end will work because what if a peak is at the edge of a chromosome. Therefore, could someone illuminate me about the positions of the chromosomes?

1) Does grch37 assembly start from 0 and continuously go until the end of mitochondrial genome ? or 2) for each chromosome it starts from 0 and end some where and for the next chromosome it starts from the beginning?

I know this is a very fundemental question but sometimes we do complicated work and forget about this basic stuff.

Thank you for the help,

Best,

Tunc.

[jotan]

Each chromosome starts at 1 and runs until the end of the chromosome. The next chromosome will begin again at 1.

For example: chr1: 1 - 197,195,432 chr2: 1 - 181,748,087

You can get chromosome sizes from UCSC. The human is here

[morovatunc]

Thank you for the answer.

Answer 1

[Devon Ryan]

bedtools slop can properly handle extending your regions at the end of chromosomes/contigs. All chromosomes and contigs start at 0 (or 1, depending on how you count).

[morovatunc]

Thank you for the answer. I think this counting will differ based on hg19 and grch37 ? right? I mean the starting from 0 or 1.

[Devon Ryan]

It only differs between file types (e.g., BED files start at 0 and GTF files at 1). The base 1 in a fasta file is always the first one, regardless of whether it's hg19 or GRCh37, which are essentially the same thing (there might be difference in the mitochondrial sequence and at least UCSC's version of hg19 merged a bunch of contigs together into *_random supercontigs).

[morovatunc]

Thank you for the response.

[morovatunc]

Dear Devon,

I am really confused with the annotations variances. Even though I read the following article. Hg19 Versus Grch37

My data is based on;

Genome Build GRCh37 Reference Name hs37d5

I am really confused at this subject. For example, bedtools have mysql connection to hg19 data. But as I checked the difference between my file regions, the only difference seems to be the "chr" part of the chr. and the annotation of unplaced contigs ("GL0190" to 'chr19_gl190' i just made it up). Other than that the start and end are same.

Could you give me some resources to at least read to make me able to see the whole picture.

Best,

Tunc.

[Devon Ryan]

There's nothing really to read on this. Different sources use different chromosome names, as far as the sequence goes that's the only big difference. If you need to know how the names relate, I have the mappings here. hg19 is UCSC, GRCh37 is Ensembl/Gencode.

[Devon Ryan]

BTW, you're using a modified version of hg19 from the 1000 genomes project. I think Heng Li has a blog post or two on the decoy sequences in hs37d5.