Expression level of mutant genes in RNAseq data

Question

[Yi]

Hello,

I have WES data from matched tumor and normal samples and mutants called from these data (in MAF files). From my understanding, if I sequence the tumor sample RNA, and run a routine RNAseq data analysis pipeline, the counts I get will be the mix of WT and MT genes. How can I get the expression (TPM) of some MT genes from tumor sample RNAseq data?

Thank you for your suggestions!

[iraun]

Hi! One way of estimating mutant allele expression this could be to count the number of RNA-Seq reads in the variant position supporting the mutation, and the total number of reads in the given position. Then, you multiply the TPM (which as you said, it is probably a mix of WT and MT counts) by the number of reads supporting the mutation, and divide it by the total number of reads in the position. You could also attempt to reconstruct the mutated allele sequence and align the RNA-Seq reads against it.

Answer 1

[i.sudbery]

This entirely depends on what fraction of your tumour cells carry the mutation. If most of the cells carry the mutation, then the expression you measure from RNAseq will be approximately the expression of the mutant.

If only a fraction of the cells carry the mutation, then this is much harder.

You might try adding the sequence of the mutant copy of the gene to the a reference transcripome FASTA, and then quantifying with Salmon. This would hoepfully give you an estimate of the fraction of copies of the that transcript in the sample that were carrying the mutation vs the fraction of copies of that transcript that were wildtype. However, that would be the balance in the sample, not what the expression level was in mutant cells vs wildtype cells.

If (and only if) your WES are from the same tumour, you could try scaling the relative expression of WT/MT from above by the fraction of cells that carry the mutation from the WES.