statistical.test.R

### Statisticla Test

#kruskal.test()

#if(p value is significant) 
  
 # do(wilcox test, but only with more than 
     
lungDiagnosisExpressionKruskal <- make.kruskal.frame(lungMetaExpression, 13, "primary_diagnosis")
x <- rownames(subset(lungDiagnosisExpressionKruskal, lungDiagnosisExpressionKruskal[1] < 0.05)) #filter significant data
x1 <- pairwise.wilcox.test(lungMetaExpression[[x[1]]]$expression, g = lungMetaExpression[[x[1]]][["primary_diagnosis"]],  p.adj = "bonf")$p.value %>% data.frame()
x2 <- x1 %>% tibble::rownames_to_column('name') %>% filter_all(. , any_vars(. < 0.05))  %>% tibble::column_to_rownames('name')   #filter and keep rownames          
x3 <- data.frame(t(x2)) %>%  tibble::rownames_to_column('name') %>% filter_all(., any_vars(. < 0.05)) %>% tibble::column_to_rownames('name') #transpose to filter out rows without significant values, cant filter out colums that easy     

wilcox_test(Value ~ factor(Group), data=data[data$Group==1|data$Group==2,])
xWil <- wilcox.test(lungMetaExpression[[x[1]]]$expression ~ lungMetaExpression[[x[1]]]$primary_diagnosis == c("Adenocarcinoma, NOS" , "Squamous cell carninoma, NOS"))
#for every row: do wilcox test if p.value is <0.05 and not the same name ... furthermore try coin::wilcox_test

rownames(x3)

#what if i subset expression values byy diagnosis factor
primaryDSpread <- lungMetaExpression$AL671277.1 %>% select(. , c(case_id, primary_diagnosis, expression)) %>% tidyr::spread(., primary_diagnosis, expression)
wilcox.test(primaryD$`Papillary adenocarcinoma, NOS`, primaryD$`Papillary adenocarcinoma, NOS`)
primaryD <- lungMetaExpression$AL671277.1 %>% select(. , primary_diagnosis, expression)
firstTest <- filter(primaryD, primary_diagnosis %in% c("Adenocarcinoma, NOS", "Squamous cell carcinoma, NOS")) #filter if they are named like this

coin::wilcox_test(formula = expression ~ primary_diagnosis, data = firstTest)
# FUNKTIONIERT