sQTL analysis about leafcutter HOT 5 CLOSED

We are working on making the scripts available. I would suggest you follow the procedure as described in the LeafCutter preprint (http://biorxiv.org/content/early/2016/03/16/044107) to qqnorm+standardize your intron ratio data, and use fastQTL(http://fastqtl.sourceforge.net/) for QTL mapping.

from leafcutter.

Thank you for your reply. There is a lot of analysis described in paragraph 6.1 of your paper so I was hoping I wouldn't have to re-create it. Any idea of when those scripts might become available? One more question, why did you decide to remove "all associations that might be caused by SNPs that overlap junction reads"? Is that because these SNPs are not sQTLs as such but mutations that happen to introduce splice donor or acceptor sites? Finally, I've come across sQTLseekeR (http://www.nature.com/articles/ncomms5698#supplementary-information) which seems like it does a similar analysis. Do you have any comments on how that compares to your analysis? Thank you very much for your time.

from leafcutter.

I'll try to make those scripts available right after US Thanksgiving. In
general, we would not remove associations where a SNP overlaps junction
reads if WASP was used to re-map reads, but we did not re-align every bam
file. In order to do a proper analysis, you must use WASP or similar
methods to remove all allelic reads that might map incorrectly or not at
all. This is time consuming and one of the reasons why I did not initially
release code for sQTL mapping: it would produce false positives if WASP was
not used.

If I recall correctly, sQTLseekeR tries to find novel isoforms, but
estimates isoform level changes, which is known to be noisy. It finds many
fewer sQTLs compared to most other methods.

Yang

On Wed, Nov 23, 2016 at 6:25 AM, Stathis [email protected] wrote:

Thank you for your reply. There is a lot of analysis described in
paragraph 6.1 of your paper so I was hoping I wouldn't have to re-create
it. Any idea of when those scripts might become available? One more
question, why did you decide to remove "all associations that might be
caused by SNPs that overlap junction reads"? Is that because these SNPs are
not sQTLs as such but mutations that happen to introduce splice donor or
acceptor sites? Finally, I've come across sQTLseekeR (
http://www.nature.com/articles/ncomms5698#supplementary-information)
which seems like it does a similar analysis. Do you have any comments on
how that compares to your analysis? Thank you very much for your time.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
#15 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AE1pieBMv6VfCxNZUwW05eSA-ey-leB7ks5rBCLAgaJpZM4K5rDB
.

from leafcutter.

Ok thank you for your advice. I'll give WASP a try in the meantime.

from leafcutter.

We've added scripts/prepare_phenotype_table.py to process the intron counts into an appropriate format for FastQTL, including splicing PCs. There's some documentation at the end of the main README too.

from leafcutter.