Zygosity annotation of mutations should be done with hisens rather than purity

[anoronh4]

We are currently using the purity facets result to annotate the maf, but I believe we should be using the hisens result instead. looks like the clinical impact runs are run using hisens, and for annotating point mutations i think it makes sense to use a result with less smoothing.

somatic annotation:

tempo/pipeline.nf

Lines 1838 to 1844 in 41a0bf5

	Rscript --no-init-file /usr/bin/facets-suite/annotate-maf-wrapper.R \
	--facets-output ${purity_rdata} \
	--maf-file ${maf} \
	--facets-algorithm em \
	--output ${outputPrefix}.facets.maf
	Rscript --no-init-file /usr/bin/annotate-with-zygosity-somatic.R ${outputPrefix}.facets.maf ${outputPrefix}.facets.zygosity.maf

germline annotation:

tempo/pipeline.nf

Lines 2335 to 2340 in 41a0bf5

	Rscript --no-init-file /usr/bin/facets-suite/annotate-maf-wrapper.R \
	--facets-output ${purity_rdata} \
	--maf-file ${maf} \
	--output ${outputPrefix}.facets.maf
	Rscript --no-init-file /usr/bin/annotate-with-zygosity-germline.R ${outputPrefix}.facets.maf ${outputPrefix}.final.maf

Here is an example annotation for a variant (select columns):

Hugo_Symbol	t_var_freq	n_var_freq	tcn	lcn	cf	purity	expected_alt_copies	allelic_imbalance	loss_of_heterozygosity	zygosity_flag
BRCA2	0.69811320754717	0.541666666666667	2	1	1	0.521339308396323	3	ALT_GAIN	TRUE	AI_LOH_ALT

We found that this led to tcn and lcn annotations in the maf that we didn't expect, for example in the following variant, tcn/lcn was 2/1, but the zygosity flag said that there was LOH. Taken together, purity and t_var_freq would suggest that it is in fact LOH. The hisens copy number says that the segment would be 1/0, and would be harmonious with t_var_freq (if i understand it correctly).

[anoronh4]

Should be noted in release notes for v2.