SNP calling using cancer cell lines

Question

[JJ]

Hi all,

I have two questions concerning SNP calling using cancer cell lines:

Comparing two human genomes on average 1 in 1000 to 1300 nucleotides is altered. Does anymore have a similar number for cancer cell lines? I was wondering if anyone knows a study where multiple cell lines were compared - I am particularly interested in the number of common and unique SNPs between them.

Concerning the SNP calling process itself, what would you recommend to get all SNPs? Call with GATK HaplotypeCaller and Mutect2 in tumour-only mode? Then merge both results?

Thanks

[ATpoint]

The question is what you want to do with the data. The point with cell lines is that they are typically in culture for decades and accumulate a plethora of genetic aberrantions over time which do not have something to do with the original cancer event. Also, as matched-normals are lacking, you cannot distinguish between somatic and germline events. What is the goal of your study?

[JJ]

Yes of course - I am aware of that. It's more for me as a 'sanity check'. We have preliminary SNP calling results for a few related cell lines, which we want to compare. Somatic and germline - all SNPs. I am just looking for some values as comparison. In particular how rapid new unique SNPs can appear in cancer cell lines when they are cultured. Thanks for input/advice in advance!

Answer 1

[Sean Davis]

Generally, cancer has a mutation rate that is small relative to natural variation. Cancer cell lines, then, have a similar total number of variants relative to the reference genome as non-cancer lines, with cancer-specific mutations being a very small proportion of total variants.

As ATPoint pointed out in the comments, it really isn't easy to make recommendations about variant calling without knowing the experimental design and the questions to answer, but in general for cancer cell lines the approach is to call variants and then "subtract" out noise and common, non-cancer-associated variations.

[JJ]

Thank you so much for your input. Do you have any idea how rapid new SNPs (not cancer-specific - any) can appear in cancer cell lines when they are cultured? If I compare a cell line before and after I have cultured it for quite a while for example, how many new SNPs could I expect?

[JJ]

Generally, cancer has a mutation rate that is small relative to natural variation. Cancer cell lines, then, have a similar total number of variants relative to the reference genome as non-cancer lines, with cancer-specific mutations being a very small proportion of total variants.

Yes, but I am interested in all SNPs - so not only cancer-specific ones. Cell lines have a much higher cell devision and growth rate than normal tissue, so eventually they would accumulate more SNPs (natural variation and cancer-specific variation). I am interested in how much more - e.g. how many SNPs can appear per passage? What it's more, cancer cell lines also have lots and lots of rearrangements within a cell population - so wouldn't call that homogenous. Hence my question is also how homogenous are the SNPs/Indels? And how does this affect SNP calling.