Finalizing Part A: Chi-Square tests and fisher tests

.~lock.lab_gwas_stats.docx#

@@ -1 +1 @@
-,noah,NovaArchSys,20.10.2022 16:12,file:///home/noah/.config/libreoffice/4;
+,noah,NovaArchSys,20.10.2022 16:40,file:///home/noah/.config/libreoffice/4;

Schrick-Noah_CS-6643_Lab-6.R

@@ -22,3 +22,35 @@ colnames(genotypes.df) <- snp.ids
 # observed contingency table for SNP rs630969
 table(phenotype,genotypes.df$rs630969,
       dnn=c("phenotype","genotype")) # dnn dimension names of table 
+dim(genotypes.df)
+
+#### Part A: Chi-Square Test
+# creates list of observed contingency tables for all SNPs
+# sapply acts on each column of genotypes.df
+observed.tables.list <- sapply(genotypes.df, function(x) 
+  table(phenotype,x,dnn=c("phenotype","genotype")))
+
+test.table <- observed.tables.list$rs634228
+genoMarg.vec  <-  colSums(test.table)           # margin vector
+phenoMarg.vec <- rowSums(test.table)            # margin vector        
+totalSubj <- sum(genoMarg.vec)                  # total subjects
+expect.test <- outer(phenoMarg.vec,genoMarg.vec/totalSubj,'*')
+
+## Fisher Test
+# Fisher exact test (chi-square test) for all SNPs
+fish_fn <- function(i){  
+  cbind(snp.ids[i], fisher.test(observed.tables.list[[i]])$p.value)    
+}
+
+# apply fisher exact test to all SNPs
+fish.df <- data.frame(t(sapply(1:ncol(genotypes.df), fish_fn)))
+colnames(fish.df) <- c("rs", "p_value")
+
+# sort SNPs by Fisher exact p-value
+if (!require("dplyr")) install.packages("dplyr")
+library(dplyr)
+fish.results <- fish.df                            %>% 
+  mutate_at("p_value", as.character) %>%
+  mutate_at("p_value", as.numeric)   %>%
+  arrange(p_value) 
+print(fish.results)