sciClone: usage for inferring tumor clonality

Question

[mp85]

Hi, I'm trying to use sciClone to infer the clonal evolution of some tumor data. I have a couple of questions about its usage (hope I'm not repeating any older question, I checked carefully but none seems satisfying):

1. I have vaf files (from the VarScan variant caller) containing both normal and tumor reads. Does it mean I can create 2 sciClone sample files (1 normal, 1 tumor) from them or should I just use tumor data?

2. What is the meaning of the five columns of the exclude regions file? (I can guess the first three are chromosome, start and end position, not quite sure about the other two)

3. To create the exclude regions file, I can use copy number data from the ICGC releases, where each variant (row) in the file is annotated as "copy neutral LOH", "gain" or "loss". I'm guessing I can populate the exclude regions file with all the rows in the file that are annotated as "copy neutral LOH", is it correct?

I apologize if I'm asking silly questions, I have not much experience in bioinformatics.

Answer 1

[Chris Miller]

1) You should be using somatic calls, which means that you should only have one output file. (in your example, created from comparing the tumor and normal using varscan's somatic mode)

2) only three columns are required for the exclude file (chr, start, stop), and that's all that sciClone will look at. One common usage is to call segmented regions of LOH, in which case there may be trailing columns giving info on those segments.

3) if you have copy-number data, you really should be using the segmented copy number data as input if available (which will give the copy number for each segment (which might be not be simple integers of gain or loss, but might be 2.15, or 3.16 etc). If all you have available is the gain or loss designations, then yeah, I guess you could put those into the exclude file. It's not ideal, though.