How Many Mutations In Wtss

Question

[Stephan]

I'm a novel student bioinformatics and I am working on a transcriptome dataset of 4 cancer patients.

I've read a lot of articles about RNA-Seq. But none of the articles explains how many mutations you typically find in Rna sequencing of humans. I have a list of 5.000 mutations that occur in all the 4 patients at same position(and more that only occurs in 1 or 2 patients) they spand 2000 known genes. dataset is alreay filtered for know Snp's Isn't that too much?? How many would you typically find ? Globally ? can't see the wood for trees

Answer 1

[David Quigley]

Daniel's points are all very good. I suspect Stephan doesn't have the raw data, but instead only the mutation summary posted by the group that published the data. In that case, you're somewhat at the mercy of the original analyst.

To address "how many mutations should I expect to find in a tumor", the answer is partly (as always), "what kind of tumor?". For example, it was recently published that lung tumors which result from long term exposure to cigarette smoke are packed full of somatic mutations. Disease which involves breaking a mismatch repair gene is likely to show many mutations or aneuploidy. I don't think we've sequenced enough tumors at this point to make informed guesses about what might be "too many" somatic alterations to believe.

[User 59]

Ah I had almost entirely neglected the physiological source of the material there, how embarassing!

[Stephan]

Thanks, nice catch. you're right I did not make the list on my own, but we are all first timers here at my department with RNA seq

[Stephan]

ohw, it's skin cancer btw

Answer 2

[User 59]

The 1000 Genomes Project should have some good data on this. Have a look at this paper. Not all genomes are full genomes in this project, some are 'exon only' and should provide some figures for your RNA-Seq work.

However, depending on your choice of pipeline, read depth and quality of your mapping etc. you're going to find this will vary the number and quality of SNPs you find. Are you confident you have been stringent enough in the data generation phase? The papers supplementary material also has information on the variant calling pipelines used for the data. You might find some of this interesting.

Answer 3

[Sheila]

I just saw a fascinating talk which among other things compared somatic point mutation rates across different cancer types, and melanoma and lung cancers have the highest rates -- between 1 and 100 mutations per Mb (this was an average value computed per tumor/patient). I'm not sure if this was derived from whole-genome or exome-capture sequencing, but thought it might be useful information for you.