methods for variant detection on pooled samples

[dnusol]

Hi,

anyone knows of methods and/or articles that may help detection of SNPs and other variants in pooled sample experiments?

Cheers,

Dave

[nickloman]

Without wanting to sound rude, you will be a much more effective scientist if you learn to use Google, Google Scholar and PubMed rather than starting your search by posting such a question here. There are plenty of references out there if you search, e.g. for "variant detection in pooled samples". If you've already searched, perhaps show what you found already and say why those methods are not useful for your application.

[Bukowski]

Try Varscan: http://bioinformatics.oxfordjournals.../17/2283.short

[nickloman]

Ah, Bukowski is kinder than me.

I should have linked to this excellent article in PloS Comp Biol:
"Ten Simple Rules for Getting Help from Online Scientific Communities"

http://www.ploscompbiol.org/article/info%3Adoi%2F10.1371%2Fjournal.pcbi.1002202

[brentp]

I have had the best results by far with Syzygy from the Broad: http://www.broadinstitute.org/software/syzygy/

It has a bunch of checks to give you good calls from pooled sample. The output
is a bit tough to deal with but it does have the best calls.


SNVer: http://snver.sourceforge.net/ also has a version for calling on pooled samples,
but it doesn't have the level of checks that Syzygy has so you'll have to do your own
filtering.

I tried a number of other methods with lower specificity and sensitivity -- testing
against genotyped data.

[dnusol]

Dear all, thanks for your kind answers. Nick, you are right, I should have commented more on my own research before posting.

I have read about a few methods, namely Varscan, CRISP, and one article by Nirajan et al, Genome Biology 2011 based on SRFIM and SERVIC4E algorithms. This article also mentions comparison of this method versus Syzygy.

My worry is that after having read a few papers and webpages found on google, I read somewhere that one method does not work well with pools made of less than a certain number of samples (number which can't recall), and another method needed comparison of pooled groups. I am a bad scientist because I did not write down these notes so I am lost there. If anyone has any comments on this... I am working with two RNA-seq pooled groups made of 30 individuals.

I will check on all the methods suggested.

All the best

Dave

[Heisman]

You may also want to look into SPLINTER, which requires a negative control to model a run's error rate and a positive control to help maximize sensitivity and specificity.

[brentp]

SPLINTER also has the distinction of being the most difficult software to download.

[kopi-o]

And Syzygy can be very hard to install, at least when I tried a few months ago ...

(EDIT: I just downloaded the latest version out of curiosity and it seemed to install cleanly now.)

Some other pooled variant callers are FreeBayes and vipR.

lh3:s list in his reply here (http://biostar.stackexchange.com/que...notype-callers) collects a number of pooled (and other) variant callers.

[PeteH]

And Syzygy can be very hard to install, at least when I tried a few months ago ...

(EDIT: I just downloaded the latest version out of curiosity and it seemed to install cleanly now.)

Good to know. We also had real difficulties trying to install Syzygy (in fact we gave up), so we'll try the latest version.

[nbafrank]

Hi!

I am actually the creator of SPLINTER! What problems did you have in downloading it? Let me know so I can potentially improve this!

Thanks!

[brentp]

Hi!

I am actually the creator of SPLINTER! What problems did you have in downloading it? Let me know so I can potentially improve this!

Thanks!

@nbafrank, where's the link so one can wget a tar ball without registering?

[zhiwei]

@brentp
You may simply prioritize the candidate variants by their output p values and choose the cutoff that you feel appropriate. It is good that you can give us feedback of the problem you have for improving this software.

I have had the best results by far with Syzygy from the Broad: http://www.broadinstitute.org/software/syzygy/

It has a bunch of checks to give you good calls from pooled sample. The output
is a bit tough to deal with but it does have the best calls.


SNVer: http://snver.sourceforge.net/ also has a version for calling on pooled samples,
but it doesn't have the level of checks that Syzygy has so you'll have to do your own
filtering.

I tried a number of other methods with lower specificity and sensitivity -- testing
against genotyped data.

[biomaseq]

download SPLINTER

Hi!

I am actually the creator of SPLINTER! What problems did you have in downloading it? Let me know so I can potentially improve this!

Thanks!

I want to use SPLINTER from http://genome.cshlp.org/content/earl...7.110.abstract.

First, download SPLINTER from https://www.ibridgenetwork.org/user/..._confirm/58621. But I cannot find the download option. Where can I get the sorftware？ Could you please help me for the problem?
Thank you！

[biomaseq]

syzygy install

And Syzygy can be very hard to install, at least when I tried a few months ago ...

(EDIT: I just downloaded the latest version out of curiosity and it seemed to install cleanly now.)

Some other pooled variant callers are FreeBayes and vipR.

lh3:s list in his reply here (http://biostar.stackexchange.com/que...notype-callers) collects a number of pooled (and other) variant callers.

Hi,

I also want to use the sorftware to do something. when i install ,much problems!

I install it according to the userguide.

when I run “syzygy -h”, some errors

——————————————————————————————
File "/bin/SAMpileuphelper.py",line 28, in <module>
from statslib import binomial,mean
ImportError: No module named statslib
_______________________________________________

how solves the problem