Concern On Tumor Samples With Mixed Cell Population

Question

[waihungh]

I'm working on cancer exome and transcriptome sequencing project. Sequencing data on tumor samples are derived from a mixed population of cells. Is there any good way to handle this issue?

[fo3c]

Do you have data from a normal (i.e. not tumour) control?

[waihungh]

No, just tumour samples

Answer 1

[Irsan]

Good point! You can purify tumor samples and seperate tumor cells from stroma and immune cells by fluorescence activated cell sorting (FACS) prior to you genomics experiment. You can do this on both fresh-frozen tissue AND on formalin fixed embedded tissue. People usually think that FACS on formalin fixed embedded material does not work but actually it does! See papers from Wim Corvers at the LUMC:

• Facs on FFPE samples for tumor purification
• Facs on human solid tumors

Answer 2

[Erik Garrison]

You can also deal with this using a pooled detection approach. freeabyes is capable of this. See: https://groups.google.com/d/topic/freebayes/Q-TFF8ollC4/discussion. In short, you would align as usual and then call using a pipeline of this form:

freebayes -f ref.fa --pooled-discrete --pooled-continuous alignments.bam

The AO (alternate observation count) and RO for each loci are then sufficient to estimate the allele frequency across your heterogenous sample.

For post-analysis you can use clustering algorithms on the observation frequency information to classify calls according to subclone. I'm not familiar with what methods are available to do this.