Remove related prior to PCA and HWE

[jfear]

The legacy workflow prunes sets of subjects using IBS/IBD and Call Rate [1]. These related subjects are removed prior to running PCA [2] and HWE [3]. There is some discussion about doing this in different ways (#10) but here I just want to replicate the legacy process. Unfortunately, testing is going to be difficult because the current test data does not contain any related sample/subject (#2).

Action Items

• Create a list of related subjects using IBS/IBD and Call Rate (i.e., refactor [1])
• Create a rule to filter these subjects and plug them into the PCA (MAF+LD) and HWE (MAF+Autosome+SNPs) filter process.

Links

• [1]

  GwasQcPipeline/modules/Snakefile_remove_related

  Lines 3 to 10 in 1d15a35

  	rule make_related_list:
  	input:
  	track = 'subject_level/SampleUsedforSubject.csv',
  	ibd = 'ibd/samples.genome',
  	fam = 'subject_level/subjects.fam',
  	imiss = 'subject_level/subjects_qc.imiss'
  	output:
  	'remove_related/subjects_to_remove.txt'

• [2]

  GwasQcPipeline/modules/Snakefile_pca

  Lines 28 to 34 in 1d15a35

  	rule extract_ld_prune_pca:
  	input:
  	bed = 'split_by_pop/{pop}_subjects.bed',
  	bim = 'split_by_pop/{pop}_subjects.bim',
  	fam = 'split_by_pop/{pop}_subjects.fam',
  	prune = 'pca/{pop}_ldPruneList.prune.in',
  	related = 'remove_related/subjects_to_remove.txt'

• [3]

  GwasQcPipeline/modules/Snakefile_HWP

  Lines 33 to 39 in 1d15a35

  	rule subset_controls:
  	input:
  	bed = 'split_by_pop/{pop}_subjects.bed',
  	bim = 'split_by_pop/{pop}_subjects.bim',
  	fam = 'split_by_pop/{pop}_subjects.fam',
  	keep = 'HWP/{pop}_controls.txt',
  	related = 'remove_related/subjects_to_remove.txt'

Related

• Add Unexpected Replicate Filter to subject level workflow #6 Add new columns to qc_summary to simplify workflow (i.e., stop making some intermediate outputs?)
• Refactor PLINK pruning rules #9 Generalize the plink pruning
• Change handling of related subjects for PCA and HWE #10 Change what we do for PCA and HWE