from gene coverage to sex determination

[paumarc]

Hello everybody

I am interested in determining sex from my nextseq runs. I read somewhere that a possibility is to calculate gene coverage for X and Y genes and then do sex determination

if cov(x)> cov(y) -> XX
if cov(x)~=cov(y) -> XY

to determine X coverage in Linux I am using samtools

samtools view <file.bam> RNAME 'chr[XY]'|wc -l

and the data I'm getting looks like

paumarc@bioinf[refSamples] samtools view wtsorted.bam RNAME 'chrY'|wc -l [ 4:39]
[main_samview] region "RNAME" specifies an unknown reference name. Continue anyway.
93504
paumarc@bioinf[refSamples] samtools view wtsorted.bam RNAME 'chrX'|wc -l [ 4:39]
[main_samview] region "RNAME" specifies an unknown reference name. Continue anyway.
1020263

The question is: Can I determinate sex only with this or i should do some kind of correction? it is normal to have more reads on chrx as long as it is bigger)

thanks

pau

[dpryan]

You'll definitely need to correct for the difference in size (Y is much smaller than X), but they may have differences in mappability as well. I would encourage you to look at a few samples for which you know the gender and that'll give you some decent numbers to start with.

BTW, "samtools view -c foo.bam chrX" will be faster than piping to "wc -l".

[Michael.Ante]

With samtools idxstats you'll get the result even faster plus the length of the chromosomes.

Since you may get mapping artefacts from intron regions, you can use a bed-file of known (chrY / chrX) - genes to limit your look-up:

samtools view -c -L chrX-genes.bed foo.bam 
samtools view -c -L chrY-genes.bed foo.bam

The result should be corrected, as Devon posted.

BTW.: in your command you use the keyword RNAME; this is not necassary.