ciangene

Order of scripts:

first.step.R

  Takes the snpStats files from the data release and turn into a numeric matrix
  Establishes UCLex External Controls (Uex) and calculates their summary stats
  Creates ReadDepth.sp

variant_filtering/qc.sh

  filters by Uex missingness and MAF
  Makes geno bed file

variant_filtering/filter_snps.R
```
  filters by function - gene tests
```

PCA

  Identify UCLex sample ancestry by comparing to 1000G. Remove non Caucasians from further analyses.

new_make_phenotype_file.R

  Creates a large matrix of phenotypes, grouped by names basically.
  Fix sample groupings
  For each phenotype, create external control set. Use for QC and for MAF filter. 
  For each phenotype, remove confounding controls.

Case control missingness

  For every selected phenotype, identify variants that have a significantly different missingness rate between cases and controls. Useful for subsequent QC.

convert_genotype_to_missingNonMissing.sh

  turns NA/non NA into 0-1 for LDAK/plink use
  Makes TK bed

make_kinships_new.sh

  Makes technical kinship (TK), Read Depth (RD) and geno kinship matrix
  calculate and plot PCs

check_tk_residuals.sh

  Checks how much kinships explain for all phenos. 
  Creates new pheno files, using residuals from (B) as phenos

LDAK/run_ldak_on_all_phenos.sh & run_ldak_on_all_phenos_res.sh

  run_ldak_on_all_phenos.sh - gene tests - all UCLex phenos, with and without kinship
  run_ldak_on_all_phenos_res.sh - gene tests using new pheno file with TK residuals.

plot_gene_test.R
```
  plot gene based results from LDAK. 
```

plink_single_variant_tests.sh

  single variant tests - Fisher, logistic regression with and without techPC covariates

run_fastLMM_on_all_phenotypes.sh

  first, run prepare_kinship_matrix_for_fastLMM.R

  single variant tests - FastLMM - Linear Regression / Mixed model with and without techKin

merge_gene_tests.sh

  Doesn't just merge. Also uses permuted pvalue in regress1..regressN to define null distribution (permuted pheno with kin) to calc final pvalue in regressALL. Script hangs when run on just one chunk/gene (eg as per power studies).

plot_singleVariant_results.R

  Merges single variant tests and plots some of them.

annotate_qqplot.R

  Makes qqPlots of snpStats style but allows you to label points too.

CianMurphy/ciangene

ciangene

Order of scripts: