SCAIP-genetic

1/6/2021 JR
This directory /wsu/home/groups/piquelab/SCAIP/SCAIP1-6_protein-coding contains all the scripts and results of SCAIP1-6 genetic analyses on protein-coding genes only

eQTL mapping

./eQTL - all the scripts and results of eQTL mapping (and follow-up analyses) on SCAIP1-6 pseudo-bulk GE data
strategy: FastQTL on pseudo-bulk residuals
INPUT: /nfs/rprdata/julong/SCAIP/analyses/SCAIP-B1-6_2020.03.23/6_DEG.CelltypeNew_output/Filter2/YtX_sel.comb.RData
OUTPUT: ./eQTL/eQTL_output/*.GEPC0.nominals.eQTL.txt.gz

Filters:

• inheritted from /nfs/rprdata/julong/SCAIP/analyses/SCAIP-B1-6_2020.03.23/6_DEG.CelltypeNew.R
  
• eQTL mapping window: +/-50kb
  
• MAF: >=10% in cohort
• gene: 0.1 CPM in more than 20% of the samples
  

Steps:

1. Normalize GE data and extract residuals: ./eQTL/normalize-all.R
2. Calculate GE PCs (to correct for in eQTL mapping): ./eQTL/GE_PCA.sh
3. Run eQTL mapping: ./eQTL/run.FastQTL.nominals-all-covs.sh
4. Check removing how many GE PCs generates the most egenes: ./eQTL/process-nominals.sh
5. Get egenes and eQTL coordinates from optimal model results: ./eQTL/get-egenes-eQTLs.sh

mashr_eQTL and mashr_dispersionQTL and mashr_eQTL-on-mean

./mashr_eQTL - all the scripts and results of running mashr (and follow-up analyses) on SCAIP1-6; relies on output in ./eQTL
strategy: mashr on eQTL results from all treatments and conditions at once
INPUT: ./eQTL/eQTL_output/.GEPC0.nominals.eQTL.txt.gz
OUTPUT: ./mashr_eQTL/output/

Filters:

• gene-variant pairs with estimates across all cell types and conditions

Steps:

1. Generate the files needed for mashr (SEs, pvalues, lfsr): ./mashr_eQTL/mashr_prep.sh
2. Run mashr on full eQTL results and save the model fit: ./mashr_eQTL/mashr-fit-model.R
3. Run mashr on chunked data using pre-saved model fit: ./mashr_eQTL/mashr_compute_posterior_chunks.sh
4. Make upset plot of mashr results across conditions: ./mashr_eQTL/plot_upset_mashr.R

eQTL mapinng on mean GE

./eQTL_mapinng-on-mean - all the scripts and results of eQTL mapping (and follow-up analyses) on SCAIP1-6 mean values from NB model strategy: FastQTL on quantile-normalized mean values from NB model while optimizing number of GE residuals PCs to remove
INPUT: /nfs/rprdata/julong/SCAIP/analyses/SCAIP-B1-6_2020.03.23/10_RNA.Variance_output/tmp9/1.2_Sel.Bx.RData
OUTPUT: ./eQTL_mapinng-on-mean/eQTL_output/*.GEPC0.nominals.eQTL.txt.gz

Filters:

• inheritted from /nfs/rprdata/julong/SCAIP/analyses/SCAIP-B1-6_2020.03.23/10_RNA.variance.R
  
• min. 3 individuals per batch-condition
  
• eQTL mapping window: +/-50kb
  
• MAF: >=10% in cohort

Steps:

1. Quantile-normalize GE measure and calculate PCs on its residuals: ./eQTL_mappinng-on-mean/normalize-all.R
2. Run dispersion-eQTL mapping: ./eQTL_mappinng-on-mean/run.FastQTL.nominals.sh
3. Get the number of eQTLs, egenes per condition and save the egenes: ./eQTL_mappinng-on-mean/process-nominals-auto.sh

dispersion eQTL mapping

./dispersionQTL - all the scripts and results of running FastQTL (and follow-up analyses) on SCAIP1-6 dispersion data
strategy: FastQTL on quantile-normalized dispersion while optimizing number of dispersion residuals PCs to remove
INPUT: /nfs/rprdata/julong/SCAIP/analyses/SCAIP-B1-6_2020.03.23/10_RNA.Variance_output/tmp9/1.2_Sel.PhxNew.RData
OUTPUT: dispersionQTL/disp-eQTL_output/*.GEPC0.nominals.eQTL.txt.gz

Filters:

• inheritted from /nfs/rprdata/julong/SCAIP/analyses/SCAIP-B1-6_2020.03.23/10_RNA.variance.R
  
• min. 3 individuals per batch-condition
  
• eQTL mapping window: +/-50kb
  
• MAF: >=10% in cohort

Steps:

1. Quantile-normalize dispersion measure and calculate PCs on its residuals: dispersionQTL/normalize-all.R
2. Run dispersion-eQTL mapping: ./dispersionQTL/run.FastQTL.nominals.sh
3. Get the number of eQTLs, egenes per condition and save the egenes: ./dispersionQTL/process-nominals.sh