Implement Tumor-Normal pair somatic variant calling

[mfoll]

We should add a feature for calling somatic mutations from multiple pairs of tumor-normal.

The basic idea will be to calculate for each normal sample the qvalue we would have if there were a mutation with AF=20%. We call that QVAL_20PC and add it in the genotype filed for all samples (put "." for tumors). Using this, we can classify mutations in the tumor as SOMATIC, GERMLINE or UNKNOWN and write this as a new genotype field (SOMATIC_STATUS):

• QVAL_TUMOR>50 & QVAL_20PC<threshold -> UNKNOWN
• QVAL_TUMOR>50 & QVAL_20PC>threshold & QVAL_NORMAL>threshold -> GERMLINE
• QVAL_TUMOR>50 & QVAL_20PC>threshold & QVAL_NORMAL<threshold -> SOMATIC
  QVAL_20PC should always be calculated in case we change the threshold afterward.
  All normals will have "." as SOMATIC_STATUS.
  If QVAL_TUMOR<50 put "." as SOMATIC_STATUS.

EDIT: Also flag tumors when low power to detect a mutation with AF=X%

• QVAL_TUMOR<50 & QVAL_XPC_TUMOR < threshold -> ".."

[mfoll]

Things to add/change:

• new input file with two columns: tumor and normal (should match with the option used for sample names: either BAM SM field or file name). like here: https://github.com/IARCbioinfo/needlestack/blob/dev/needlestack.nf#L211
• check that all sample names in this file actually exist (easy to do in R but maybe too late). Not the priority, we can do a dedicated process at the beginning.
• we allow some additional samples (not in the tumor-normal pair list) and put "." in the two additional genotype fields.
• find a way to switch to "tumor-normal" mode in R if the new input file has been provided by the user (flag? test presence of file? test file size?)
• load this file in R in the VCF creation process (new input here: https://github.com/IARCbioinfo/needlestack/blob/dev/needlestack.nf#L401)
• After the regression (here https://github.com/IARCbioinfo/needlestack/blob/dev/bin/needlestack.r) calculate the QVAL_20PC for all normal like here https://github.com/IARCbioinfo/needlestack/blob/dev/bin/plot_rob_nb.r#L82
• Add the new genotype fields (ex for SNV here https://github.com/IARCbioinfo/needlestack/blob/dev/bin/needlestack.r#L233)
• Write the QVAL_20PC and the status here https://github.com/IARCbioinfo/needlestack/blob/dev/bin/needlestack.r#L300
• Thinks about a possible option to output only positions where there is a somatic mutation
• Write QVAL_XPC in the VCF for Tumors, and add X as input
• Flag in all samples mutations detected as GERMLINE in some samples