silent and nonsilent rates are too different

We run the analysis using Mutect and annovar,after that we input the annotated data into MutSig. However, it gave the error message:

silent and nonsilent rates are too different

Any clue why MutSig spit out this error?

Thanks!

We have also got the same

We have also got the same error, going through the code it seems ratio of # nonsilent rate/ # silent rate (#nonsilent rate = # non silent/sum of coverage of all nonsilent mutations) should be between 0.5 to 2.
The error that we are getting could be due to fact we are using full exome coverage file, trying to create a coverage file our own sample.

Why are silent and non-silent rates too different?

In general, MutSig estimates the background mutation rate by looking at the silent to non-silent mutation rates, which in turn is used to calibrate the p and q values, and to correct for per-gene differences in background mutation rate.

In some cases, all the silent mutations get filtered out of the maf file before the user receives it. In other cases, if the coverage information does not reflect the areas the mutation caller used, the results can be off - eg if the coverage information claims mutations could potentially be called in intronic regions and adjacent to the gene, but in fact none were called, this would artificially reduce the silent rate.

I got the same problem. Any

I got the same problem.

Any solutions so far?? Does anyone know how to generate the coverage file from BAM file?

Thanks