Driver cancer genes

Question

[Bnf83]

Hi guys, is it possible that there exist tumor patients without mutations in driver cancer genes?

Kind regards,

B

[WouterDeCoster]

Perhaps no mutations in known driver genes, but I guess by definition there has to be a driver mutation somewhere.

[Bnf83]

Yes I think you're right. Do you know where can I find some references about?

[ivivek_ngs]

How are you defining mutations as driver status is the thing you should take into account. There are different ways to assign a driver status to mutations in a patient or patient cohort from a pool of mutation. Depends on various factors . What criteria are you looking at?

[Bnf83]

Ok I understand. I'm referring to frequently mutated genes like TP53, MYC, etc.

[ivivek_ngs]

I would in any case refer some tools that can take mutations from your dataset and try to find if they can be assigned driver within sample, cross-sample cohort of TCGA and also have functionality and structality impact and providing them a driver status. Just COSMIC genes at times might not give the driver status to your mutations. It also depends on how many samples you have in your case then one can try to construct the mutational driver model to assess. Try to see intOGen tool, if it work out for your analysis for driver assignment.

[Jean-Karim Heriche]

The COSMIC cancer gene census is a list of genes whose mutations are believed to cause cancer.

[Collin]

It has currently been assessed that we have a poor understanding of the driver genes mutated at rare or intermediate frequencies in a cohort (http://www.ncbi.nlm.nih.gov/pubmed/24390350 ). So it is likely if your drivers fall into those categories, genes wouldn't be commonly recognized. Additionally, identifying the cancer driver mutations in a sample is contingent on the sensitivity of the somatic variant calling, i.e., given a somatic mutation exists what is the probability it is detected in variant calling. If the variant calling didn't have enough sensitivity, you could have a driver mutation in say TP53, but you didn't realize it because there was no variant call.

[ivivek_ngs]

Yes that is also the reason I followed up with the tool I suggested but before the OP needs to run callers which will asses mutation not only with high frequency sensitivity like VarScan2 but also with low frequency mutations like that of Mutect and then use the vcf output from the callers to define drivers with the intogen tool. I would suggest OP also reads the paper of different mutation callers having different models of associating a high confidence call variant. Take a look at this link.

[Chris Miller]

Every tumor has to contain one or more cancer-intiating driver events at high VAF in the founding clone. Depending on what biological questions the study is asking (intiation vs progression/resistance), low-VAF mutations may have more or less importance.

[Collin]

If you only want to use pre-defined cancer driver genes, then there is a plethora of lists floating out there: Vogelstein et al., MutsigCV list, TCGA Pancancer 12, and cancer gene census.

Answer 1

[russhh]

Obviously, certain viruses can initiate oncogenesis by introducing their own drivers into the host cell (rather than mutating host-cell drivers). But also, it's very likely that solid tumours may lack mutations in known driver genes, since these gene lists are continually being redefined as more studies are performed.

... and look into medulloblastoma

... and not all mutations in driver genes are driver mutations

[Bnf83]

Is there some literature about this?

[russhh]

oooo00000000odles of it