Artifactual mutations in CCLE data

Question

[monukmr98]

Hi there,

I am confused whether some of the mutations reported in some big study like CCLE are artifactual mutations. I have been asked this question many a times, but don't understand what artifactual mutations are ?

Are artifactual mutations related to:

platform related - eg. illumina, 454 etc

protocol related - eg. library preparation for rna-seq or exome-seq purposes

variant calling algorithm related - eg GATK , SAMTOOLS mpileup

or something else what I mite be missing

Please shed some light on this topic.

Thanks

Regards

[russhh]

All of the above, plus a number of them are germline variants which weren't identified since no non-transformed sample is available for the vast majority of cell lines

[monukmr98]

Thanks russ

In their preferred dataset, CCLE people have mentioned that the following variants were filtered out:

• common polymorphisms
• allelic fraction < 10%
• putative neutral variants (missenses present in less than 2 warm-blooded vertebrates)
• located outside of the CDS for all transcripts

It means, that germline variants have not been filtered out.

Also, I want to know whether is it possible to filter out germ line variants from cell line samples or only patient samples are eligible for that ? If yes, then what are the software or pipelines available?

Thanks

[russhh]

to be honest, I don't think it would be possible to predict the germline variants for cell lines that don't have a comparator. It's hard enough in tumours, where untransformed cells are present, to determine the somatic variants from the germline variants unless you have some blood DNA or somehting