The ABSOLUTE input and HAPSEG output format

Question

[shiyang93]

HI I want to use the ABSOLUTE to analyze my WES data. I have already segmented my data, with log2 ratio of both allele(log2-allele1, log2-allele2, and log2-total) in each segment.

I noticed that there are two input formats for ABSOLUTE, one is output data from HAPSEG, the other is a TAB delimited segmentation file. However, if the second format is used, then the 'copy_num_type' argument must be set as 'total'(There are two choices: 'total' or 'allelic'). http://archive.broadinstitute.org/cancer/cga/absolute_run. I want to use 'allelic' for this argument, but then I must use the output data from HAPSEG as an input.

I think the information of my segmented data is enough for the ABSOLUTE, but I really do not know what the HAPSEG output data looks like. Could anyone who used HAPSEG show me its output format? Or who is familiar with ABSOLUTE tell me how to use the 'ALLELIC' choice to run ABSOLUTE with my segmented data?

THANKS!

Answer 1

[markus.riester]

The tools to make original ABSOLUTE work with WES data are AFAIK not publicly available. There are other tools now available that provide allele-specific copy numbers and are optimized for WES data. For example Sequenza, FACETs or, pardon the shameless advertising, our own tool PureCN.

[shiyang93]

Thanks！So that means if I have already done the segmentation work by myself and feed the results to ABSOLUTE, it still cannot work correctly?

[markus.riester]

Not sure, if your allele-specific log-ratio calculation is good, then original ABSOLUTE should work fine in theory. You can try contacting the author.

Purity/ploidy inference is a problem that quickly blows up if some upstream step (e.g. coverage normalization, segmentation, artifact filtering) is done poorly and all 3 tools provide a complete pipeline from BAM files and are optimized for WES data. To some extend, some of them also take uncertainty of various steps in the likelihood model into account.