#!/usr/bin/Rscript   
getwd()
setwd("~/git/mosaic_2020/")
getwd()
########################################################################
# TASK: check stats
########################################################################
# load libraries, packages and local imports
source("Header_TT.R")

#=============
# Input
#=============
source("data_extraction_mediators.R")
#source("read_data.R")

########################################################################
#=============
# Data extraction for non-severe
#=============
table(fp_adults_ics$T1_resp_score)
table(fp_adults_ics$obesity)

df = fp_adults_ics[fp_adults_ics$T1_resp_score<3,]
table(df$obesity)

df_ob = df[df$obesity == 1,]
df_nob = df[df$obesity == 0,]

rm(npa_lf, npa_wf, sam_lf, sam_wf, serum_lf, serum_wf, fp_adults_ics)

#============
# npa
#============
# vl_pfu_ul: t2
wilcox.test(df_ob$vl_pfu_ul_npa2, df_nob$vl_pfu_ul_npa2)
wilcox_test(vl_pfu_ul_npa2 ~ obesity, data = df)

foo = data.frame(df$obesity, df$vl_pfu_ul_npa2)
na_c = sum(is.na(foo$df.vl_pfu_ul_npa2))
nrow(df) - na_c

tapply(df$vl_pfu_ul_npa2, df$obesity, median, na.rm = T)
tapply(df$vl_pfu_ul_npa2, df$obesity, summary, na.rm = T)

# 1l29: t1
wilcox.test(df_ob$il29_npa1, df_nob$il29_npa1)

foo = data.frame(df$obesity, df$il29_npa1)
na_c = sum(is.na(foo$df.il29_npa1))
nrow(df) - na_c

tapply(df$il29_npa1, df$obesity, median, na.rm = T)
tapply(df$il29_npa1, df$obesity, summary, na.rm = T)

# tnfr1: t3
wilcox.test(df_ob$tnfr1_npa3, df_nob$tnfr1_npa3)

foo = data.frame(df$obesity, df$tnfr1_npa3)
na_c = sum(is.na(foo$df.tnfr1_npa3))
nrow(df) - na_c


tapply(df$tnfr1_npa3, df$obesity, median, na.rm = T)
tapply(df$tnfr1_npa3, df$obesity, summary, na.rm = T)

boxplot(df_ob$tnfr1_npa3, df_nob$tnfr1_npa3)

# rantes: t1
wilcox.test(df_ob$rantes_npa1, df_nob$rantes_npa1)

foo = data.frame(df$obesity, df$rantes_npa1)
na_c = sum(is.na(foo$df.rantes_npa1))
nrow(df) - na_c

#============
# sam
#============
# tarc: t3
wilcox.test(df_ob$tarc_sam3
            , df_nob$tarc_sam3)

foo = data.frame(df$obesity, df$tarc_sam3)
na_c = sum(is.na(foo$df.tarc_sam3))
nrow(df) - na_c

tapply(df$tarc_sam3, df$obesity, median, na.rm = T)


# il13: t3
wilcox.test(df_ob$il13_sam3
            , df_nob$il13_sam3)

foo = data.frame(df$obesity, df$il13_sam3)
na_c = sum(is.na(foo$df.il13_sam3))
nrow(df) - na_c

tapply(df$il13_sam3, df$obesity, median, na.rm = T)


# eotaxin: t1
wilcox.test(df_ob$eotaxin_sam1
            , df_nob$eotaxin_sam1)

foo = data.frame(df$obesity, df$eotaxin_sam1)
na_c = sum(is.na(foo$df.eotaxin_sam1))
nrow(df) - na_c

tapply(df$eotaxin_sam1, df$obesity, median, na.rm = T)


#============
# serum
#============
# ifna2a: t1
wilcox.test(df_ob$ifna2a_serum1, 
            df_nob$ifna2a_serum1)

foo = data.frame(df$obesity, df$ifna2a_serum1)
na_c = sum(is.na(foo$df.ifna2a_serum1))
nrow(df) - na_c

tapply(df$ifna2a_serum1, df$obesity, median, na.rm = T)


# mip17: t1
wilcox.test(df_ob$mip17_serum1, 
            df_nob$mip17_serum1)

foo = data.frame(df$obesity, df$mip17_serum1)
na_c = sum(is.na(foo$df.mip17_serum1))
nrow(df) - na_c

tapply(df$mip17_serum1, df$obesity, median, na.rm = T)

# il1: t1
wilcox.test(df_ob$il1_serum1
            , df_nob$il1_serum1)

foo = data.frame(df$obesity, df$il1_serum1)
na_c = sum(is.na(foo$df.il1_serum1))
nrow(df) - na_c

tapply(df$il1_serum1, df$obesity, median, na.rm = T)
tapply(df$il1_serum1, df$obesity, summary, na.rm = T)
boxplot(df_ob$il1_serum1, df_nob$il1_serum1)


# ifna2a: t2 
wilcox.test(df_ob$ifna2a_serum2, 
            df_nob$ifna2a_serum2)

foo = data.frame(df$obesity, df$ifna2a_serum2)
na_c = sum(is.na(foo$df.ifna2a_serum2))
nrow(df) - na_c


tapply(df$ifna2a_serum2, df$obesity, median, na.rm = T)

# ifna2a: t3
wilcox.test(df_ob$ifna2a_serum3, 
            df_nob$ifna2a_serum3)

foo = data.frame(df$obesity, df$ifna2a_serum3)
na_c = sum(is.na(foo$df.ifna2a_serum3))
nrow(df) - na_c

tapply(df$ifna2a_serum3, df$obesity, median, na.rm = T)


# ifnb: t3
wilcox.test(df_ob$ifnb_serum3
            , df_nob$ifnb_serum3)

foo = data.frame(df$obesity, df$ifnb_serum3)
na_c = sum(is.na(foo$df.ifnb_serum3))
nrow(df) - na_c

tapply(df$ifnb_serum3, df$obesity, median, na.rm = T)