From 18797dfeb4c437c38dd40e4505d761407d83e1d5 Mon Sep 17 00:00:00 2001
From: Gibran Hemani <g.hemani@bristol.ac.uk>
Date: Sun, 20 Oct 2024 11:08:20 +0100
Subject: [PATCH] renaming pc file for convenience

---
 01-ancestry.sh                               | 16 ++++++++--------
 03-prs.sh                                    |  3 ++-
 resources/phenotypes/organise_phenotypes.rmd |  2 +-
 test/covariates.r                            |  2 +-
 4 files changed, 12 insertions(+), 11 deletions(-)

diff --git a/01-ancestry.sh b/01-ancestry.sh
index c39dd4f..8b9dde3 100755
--- a/01-ancestry.sh
+++ b/01-ancestry.sh
@@ -205,10 +205,10 @@ then
 
     section_message "pcs"
 
-    if test -f "${genotype_processed_dir}/${bfile_prefix}pc.txt"; then
+    if test -f "${genotype_processed_dir}/pcs.txt"; then
         echo "pcafile already provided"
         Rscript resources/genotypes/genetic_outliers.r \
-            ${genotype_processed_dir}/${bfile_prefix}pc.txt \
+            ${genotype_processed_dir}/pcs.txt \
             ${env_pca_sd} \
             ${env_n_pcs} \
             ${genotype_processed_dir}/${bfile_prefix}_genetic_outliers.txt \
@@ -229,16 +229,16 @@ then
             pcs_unrelated
             pcs_related            
             sed -i 1d ${genotype_processed_dir}/scratch/fastpca_pcs_related.txt
-            cat ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/scratch/fastpca_pcs_related.txt > ${genotype_processed_dir}/${bfile_prefix}pc.txt
+            cat ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/scratch/fastpca_pcs_related.txt > ${genotype_processed_dir}/pcs.txt
 
         else
             pcs_unrelated
-            cp ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/${bfile_prefix}pc.txt
+            cp ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/pcs.txt
         fi
 
         echo "Check PCs e.g. by plotting them"
         Rscript resources/genotypes/genetic_outliers.r \
-            ${genotype_processed_dir}/${bfile_prefix}pc.txt \
+            ${genotype_processed_dir}/pcs.txt \
             ${env_pca_sd} \
             ${env_n_pcs} \
             ${genotype_processed_dir}/${bfile_prefix}_genetic_outliers.txt \
@@ -271,18 +271,18 @@ then
                 pcs_unrelated
                 pcs_related            
                 sed -i 1d ${genotype_processed_dir}/scratch/fastpca_pcs_related.txt
-                cat ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/scratch/fastpca_pcs_related.txt > ${genotype_processed_dir}/${bfile_prefix}pc.txt
+                cat ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/scratch/fastpca_pcs_related.txt > ${genotype_processed_dir}/pcs.txt
 
             else
                 pcs_unrelated
-                cp ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/${bfile_prefix}pc.txt
+                cp ${genotype_processed_dir}/scratch/fastpca_pcs_unrelated.txt ${genotype_processed_dir}/pcs.txt
 
             fi
 
             mv ${results_dir}/01/pcaplot.png ${results_dir}/01/pcaplot_round1.png
 
             Rscript resources/genotypes/genetic_outliers.r \
-                ${genotype_processed_dir}/${bfile_prefix}pc.txt \
+                ${genotype_processed_dir}/pcs.txt \
                 ${env_pca_sd} \
                 ${env_n_pcs} \
                 ${genotype_processed_dir}/${bfile_prefix}_genetic_outliers.txt \
diff --git a/03-prs.sh b/03-prs.sh
index 45acd2a..c57a394 100755
--- a/03-prs.sh
+++ b/03-prs.sh
@@ -8,6 +8,7 @@ source config.env
 
 # create results directory
 mkdir -p ${results_dir}/03
+mkdir -p ${results_dir}/03/ldmats
 
 # log everything from this script to a logfile in the results director
 exec &> >(tee ${results_dir}/03/logfile)
@@ -130,7 +131,7 @@ do
             --keep ${genotype_processed_dir}/scratch/ldmats/keeptemp \
             --bfile ${genotype_processed_dir}/scratch/tophits/${ph} \
             --r2-unphased ref-based bin4 yes-really \
-            --out ${results_dir}/03/${filename}
+            --out ${results_dir}/03/ldmats/${filename}
 
         rm -r ${genotype_processed_dir}/scratch/ldmats
     fi
diff --git a/resources/phenotypes/organise_phenotypes.rmd b/resources/phenotypes/organise_phenotypes.rmd
index 946b449..dfd96cf 100755
--- a/resources/phenotypes/organise_phenotypes.rmd
+++ b/resources/phenotypes/organise_phenotypes.rmd
@@ -68,7 +68,7 @@ os$phenotypes <- phenotypes
 
 ```{r}
 gen_covs <- read_gen_covs(
-  file = file.path(Sys.getenv("genotype_processed_dir"), paste0(Sys.getenv("bfile_prefix"), "pc.txt")),
+  file = file.path(Sys.getenv("genotype_processed_dir"), paste0("pcs.txt")),
   npcs = as.numeric(Sys.getenv("env_n_pcs"))
 )
 str(gen_covs)
diff --git a/test/covariates.r b/test/covariates.r
index bcb8a8b..678350e 100644
--- a/test/covariates.r
+++ b/test/covariates.r
@@ -5,6 +5,6 @@ library(here)
 library(dotenv)
 readRenviron(here("config.env"))
 
-a <- fread(file.path(Sys.getenv("genotype_processed_dir"), paste0(Sys.getenv("bfile_prefix"), "pc.txt")), header=TRUE)
+a <- fread(file.path(Sys.getenv("genotype_processed_dir"), paste0("pcs.txt")), header=TRUE)
 
 b <- tibble(FID=a$FID, IID=a$IID, yob = sample(1920:1980, length(FID), replace=TRUE), sex=sample(1:2, length(FID), replace=TRUE), bp_med = rbinom(length(FID), 1, 0.2), choleterol_med= rbinom(length(FID), 1, 0.2), parity=sample(1:5, length(FID), replace=TRUE)
\ No newline at end of file