setwd("/home/tanu/git/LSHTM_analysis/scripts/plotting")

source("combining_dfs_plotting.R")

table(merged_df3$mutation_info)

# assign foldx
#ddg<0 = "Stabilising" (-ve)
table(merged_df3$ddg < 0)
merged_df3$foldx_outcome = ifelse(merged_df3$ddg < 0, "Stabilising", "Destabilising")

#===========
# PS data
#===========
dr_muts = merged_df3[merged_df3$mutation_info == "dr_mutations_pyrazinamide",]
other_muts = merged_df3[merged_df3$mutation_info == "other_mutations_pyrazinamide",]

par(mfrow = c(1,1))
par(mfrow = c(2,6))

# mcsm duet
boxplot(dr_muts$duet_scaled, other_muts$duet_scaled, main = "DUET"
        #, col = factor(merged_df3$duet_outcome)
        )
wilcox.test(dr_muts$duet_scaled, other_muts$duet_scaled, paired = F)

# foldx ddg
boxplot(dr_muts$ddg, other_muts$ddg, main = "Foldx")
wilcox.test(dr_muts$ddg, other_muts$ddg, paired = F)

# rd
boxplot(dr_muts$rd_values, other_muts$rd_values, main = "RD")
wilcox.test(dr_muts$rd_values, other_muts$rd_values)

# kd
boxplot(dr_muts$kd_values, other_muts$kd_values, main = "KD")
wilcox.test(dr_muts$kd_values, other_muts$kd_values)

# asa
boxplot(dr_muts$asa, other_muts$asa, main = "ASA")
wilcox.test(dr_muts$asa, other_muts$asa)

# rsa
boxplot(dr_muts$rsa, other_muts$rsa, main = "RSA")
wilcox.test(dr_muts$rsa, other_muts$rsa)

#===================================================================
#==========
# LIG data
#==========
dr_muts_lig = merged_df3_lig[merged_df3_lig$mutation_info == "dr_mutations_pyrazinamide",]
other_muts_lig = merged_df3_lig[merged_df3_lig$mutation_info == "other_mutations_pyrazinamide",]

# mcsm ligand affinity
boxplot(dr_muts_lig$duet_scaled, other_muts_lig$duet_scaled, main = "Ligand affinity")
wilcox.test(dr_muts_lig$duet_scaled, other_muts_lig$duet_scaled, paired = F)

# rd
boxplot(dr_muts_lig$rd_values, other_muts_lig$rd_values, main = "RD")
wilcox.test(dr_muts_lig$rd_values, other_muts_lig$rd_values)

# kd
boxplot(dr_muts_lig$kd_values, other_muts_lig$kd_values, main = "KD")
wilcox.test(dr_muts_lig$kd_values, other_muts_lig$kd_values)

# asa
boxplot(dr_muts_lig$asa, other_muts_lig$asa, main = "ASA")
wilcox.test(dr_muts_lig$asa, other_muts_lig$asa)

# rsa
boxplot(dr_muts_lig$rsa, other_muts_lig$rsa, main = "RSA")
wilcox.test(dr_muts_lig$rsa, other_muts_lig$rsa)

# checking agreement b/w mcsm and foldx
cols_to_select = c("mutationinformation"
                   , "mutation_info"
                   , "duet_scaled"
                   , "ddg"
                   , "duet_outcome"
                   , "foldx_outcome")

merged_df3_short = select(merged_df3, cols_to_select)

mcsm_foldx = merged_df3_short[which(merged_df3_short$duet_outcome != merged_df3_short$foldx_outcome),]

mcsm_foldx$sign_comp = ifelse(sign(mcsm_foldx$duet_scaled)==sign(mcsm_foldx$ddg), "PASS", "FAIL")
table(mcsm_foldx$sign_comp)

# another way of checking
merged_df3$sign_comp = ifelse(sign(merged_df3$duet_scaled)==sign(merged_df3$ddg), "PASS", "FAIL")
table(merged_df3$sign_comp)

disagreement = table(merged_df3$sign_comp)[2]/nrow(merged_df3)*100
agreement = 100 - disagreement

cat("There is", agreement, "% between mcsm and foldx predictions")