RNASeq gender identification

Question

[popayekid55]

Dear all,

i have 10 human samples. i have aligned to human grch38 from ensemble using hisat. i want to the sample gender. i have checked reads for X and Y chromosome. For all the samples X reads are greater compared to Y. i refered this link. But our samples have male samples. how is this possible. i dont have the access to GATK DepthOfCoverage. what could be the alternate for this to find coverage. or is there any other way to confirm the gender.

thank you all

[Brian Bushnell]

The X chromosome is much larger and more important than the Y chromosome, so of course it will tend to yield more reads. I would guess that 1, 6, 8, and 9 are female, but you can't really say that with any confidence without calibration. And the columns are unlabeled and no methodology is given, so it's not even certain what the numbers mean.

[popayekid55]

And the columns are unlabeled and no methodology is given, so it's not even certain what the numbers mean.

1st column is sample IDs, subsequent columns are ChrX and ChrY. Number are raw read count aligned to chromosome. Analysis methodology as i said hisat was used to align. sequencing is polyA capture and data from Hiseq2000

[Brian Bushnell]

OK, thanks. So, I suggest you follow Devon's advice, as it looks like there is a clear separation between males and females; the lowest male expression being a 170x ratio, and highest (apparent) female expression being a 3964x ratio, which gives you over a factor of 20 margin.

[Satyajeet Khare]

In absence of X-chromosomal abnormalities, Xist expression should be able help mark female samples.

[popayekid55]

these are normal samples. there are no abnormalities. could understand by Xist expression

[Kevin Blighe]

From my experience, yes, looking at the expression of Xist, you'll see great differences between males and females in your RNA-seq dataset.

[GenoMax]

i have checked reads for X and Y chromosome. For all the samples X reads are greater compared to Y.

Please provide a precise metric. What kind of samples are these?

[popayekid55]

Sample_ID   ChrX    ChrY    X/Y
1   2953921 282 10474.8971631206
2   2131045 22727   93.7671052053
3   2509846 18874   132.979018756
4   1956562 19873   98.4532783173
5   2751061 22940   119.9241935484
6   2414397 609 3964.5270935961
7   3240162 18954   170.9487179487
8   2497237 299 8351.9632107023
9   2220441 1094    2029.6535648995
10  2461161 22236   110.6836211549

these human normal male female samples

[Devon Ryan]

Take the ratio of Y and X and I think you'll see that you have two groups of samples...

[popayekid55]

I have added the ratio. Are there any strict/range of cutoff to separate groups. Different threads gives different cutoffs

[Devon Ryan]

Any strict cutoff you get from someone else will have been derived from their datasets. Just look at the data, you'll be able to eye-ball two groups.