simplify and refactor

[jackhump]

to be done by Jack

• remove multi-cohort functionality - overcomplicates things

• simplify string matching: just assume that user has symlinked a bunch of gzipped VCFs and each VCF is a chromosome for that cohort

• hardcode blacklist filtering - no point in making optional

• create sample filter rule - from results of PCA for example to remove certain samples

• add in final MAF filter - pipeline should output all QC-passed variants, plus a common variants file

• look into chunking - can this be increased? ideally chunk at beginning and do every step per chunk.

• go over cluster.yaml and snakejob - ensure optimal execution across maximum number of nodes while keeping resources low

[jackhump]

• combine sample filtering with SNP filtering into a single rule

• combine all chunk-level filters into a single rule

• work with HPC to optimise I/O for writing temp files during chunk filtering.

[jackhump]

• Replace all vcftools commands with bcftools to speed up execution.