vcftools: the length of each row is not same

Question

[zwang10]

Hello! I have vcf.gz file, and I want to change it into 012 matrix. I use follow command

vcftools --gzvcf chr1.vcf.gz --out chr1 --012

Then, it outputs chr1.012, chr1.012.pos, and chr1.012.indv. But I found the length of each row of chr1.012 is not equal. And the length of rows of chr12.012 is not same as chr1.012.

[Ram]

Just to cover all bases, what are the commands you used to find

1. the length of each row in a file
2. the number of rows in a file

This is just so we are sure there was no error in the counting logic.

[zwang10]

Hello! I use bash scripting. To print the length of each row, I use

cat chr1.012 | awk '{print NF}'

To print the number of rows of chr1.012.pos, I use

wc -l chr1.012.pos

[Ram]

Did you account for anomalies in separation? Maybe a case of multiple separators at places where it's not supposed to happen? Given the 012 file has an empty value indicator (-1), maybe squeeze the separator using a tr -s before the awk?

[zwang10]

I found the length of chr1.012.pos is always much larger than the length of row of chr1.012. So the case you mentioned would not happen.

[Ram]

Maybe they use different separators?

EDIT: Scratch that - doesn't look like it; they're all tab separated. It has something to do with the actual variants then.

[zwang10]

The real name of my vcf.gz file is GAZ00001016581_1.ALSPAC.beagle.anno.csq.shapeit.20131101.vcf.gz (this is for chromsome 1). Also there is a file called _EGAZ00001016604_ALSPAC.beagle.anno.csq.shapeit.20131101.sites.vcf.gz. I do not know whether this file is useful or not.

[zwang10]

Do you have alternative tools to convert vcf.gz file into 012 matrix?

[Ram]

Not really. Sorry, I cannot help you with this now - I would have asked for a bit of the file to examine, but I am busy with my day job this week.

[zwang10]

Sure. But my file is so large. The smallest file for chromsome is chr22 (about 4.1G).

[Pierre Lindenbaum]

Hello zwang10!

It appears that your post has been cross-posted to another site: http://stackoverflow.com/questions/36256693

This is typically not recommended as it runs the risk of annoying people in both communities.

[zwang10]

Thanks for your suggestion. I deleted cross-posted one in stackoverflow.

Answer 1

[c.v.oflynn]

Hi zwang10,

did you solve this?

the first column of chr1.012 is numeric sample id, not genotype. remove this and you have the same lengths.

Ciaran