Strelka: Somatic small-variant calling workflow for matched tumor-normal samples

[ctsa]

Hello All,

Strelka is a new workflow available to call SNVs and small indels from sequencing data for matched tumor-normal samples. It is designed to detect somatic variants at lower frequencies typically encountered in tumors due to sample impurity or sub-clone variation. The workflow also provides computational efficiency appropriate for the whole genome sequencing case: requiring ~1 core-hour per 2x combined tumor normal coverage.

More information/source code available here:

https://sites.google.com/site/strelkasomaticvariantcaller

We appreciate any feedback on how these methods can be improved.

Best Regards,

-Chris Saunders

[nilshomer]

Looks like the license is fairly restrictive. Any chance of moving this to an open source license?

[Jane M]

Hello,

I could be interested in the tool that you suggest since it deals with my problematic, but as said nilshomer, there seems to be restrictions with the license, I am not able to download the sources...

Jane

[ctsa]

Hi Nils and Jane --

Thanks for highlighting this issue, I will take a look today to see what our options are wrt the source license.

-Chris

[ctsa]

I've gotten additional feedback about the source download link on some web browsers. The source download URL is:

ftp://[email protected]

Note that no password is required. In firefox it looks like a password prompt comes up anyway -- you can leave the password field blank and just hit "Ok" to enter the ftp site.

[nilshomer]

I've gotten additional feedback about the source download link on some web browsers. The source download URL is:

ftp://[email protected]

Note that no password is required. In firefox it looks like a password prompt comes up anyway -- you can leave the password field blank and just hit "Ok" to enter the ftp site.

I am able to download the source code fine, but it is the license to which I do not agree.

[ctsa]

Hi Nils --

Sorry if there's a misunderstanding, the additional ftp advice is in response to a separate conversation. As I replied above, I'm working on the license issue.

-Chris

[nilshomer]

Sorry, thanks for looking into it!

[pravee1216]

platform independent?

Good news!!

Couple of questions:

a) Does Strelka support alignment files from Roche 454/FLX sequencing reads? or Is it designed mainly for Illumina data?

b) How does it handle calls at homopolymer regions, especially for 454/MiSeq/IoT platform data? Is this tested?

Thanks in advance

Raj

[genomicist]

Why are there two "format" fields in the output?

I wonder why there are two "format" fields in the output (the last two columns of the output file) of this type: DP:FDP:SDP:SUBDP:AU:CU:GU:TU. Have been looking for explanation but all in vane.

[genomicist]

I also wonder what are the optional "extraStrelkaArguments" that are possible to specify in the configuration file. Is there a list?

Specifically, is it possible to filter the calls on variant allele frequency? My "passed" SNV list contains lots of calls that are supported by 1 or 2 reads with the alternative base, along with a couple of hundred reads with reference base. These are presumably sequencing errors.

[lethalfang]

I'm trying to use Strelka on some sequencing data we got from Solid 5500, with its BAM file aligned with LifeScope 2.5.
The LifeScope-produced BAM file seems to be incompatible with Stralka.
Does anyone know of a way to convert the BAM into something acceptable by Strelka?

Thanks in advance.

Well, it seems I was just missing an index .bam.bai file, which I created using samtools index aln.bam.
It is running now. Let's see how it goes.

[malachig]

It looks like nilshomer's initial question was never addressed. We have the same concern with using this software:

"Looks like the license is fairly restrictive. Any chance of moving this to an open source license?"

If this could be moved to open source, that would make it easier to deploy in pipelines/platforms that are themselves open source projects...

[ctsa]

Looks like I'm not getting emails for this thread. I'll try to briefly cover the existing questions but encourage you to re-post any current issue to the strelka mailing list here:

https://groups.google.com/forum/#!forum/strelka-discuss

- License:

Strelka has recently been moved to the Illumina Open Source Software License (v1). Details are on github here:

http://cloud.github.com/downloads/sequencing/licenses/General_Illumina_Open_Source_License_Template_1_Final.pdf

- Incompatible BAMs:

All known BAM restrictions are described in the FAQ here:

https://sites.google.com/site/strelkasomaticvariantcaller/home/faq#TOC-What-BAM-files-can-be-used-in-the-workflow-

- Format Fields:

All format fields are described in the VCF header, as well as on the website here:

https://sites.google.com/site/strelkasomaticvariantcaller/home/somatic-variant-output

[Jane M]

Hello,

I am using Strelka 1.0.14 on WGS data for a few days.
Strelka ran without problem on my samples.
To further filtered my list of variants, I need information about the coverage of reference and "variant" alleles in both normal and tumor samples.

As I read in the strelka discussion list, the number of reads supporting the "indel"(=variant) allele is given by the TIR column (its first field preferentially).
My problem is to find the number of reads supporting the reference allele.
Maybe I missed something, but I tried DP, TAR, DP-TOR, TAR+TOR,... without success....
In some cases, DP seem ok. In other cases, TAR+TOR look ok.
This information is crucial for my analysis and I am stuck here with this detail...

Could you please tell me how to compute precisely the number of reads supporting the reference allele?

Thank you in advance,
Jane

ps: I just posted this question on the strelka mailing list, but it could be more visible here