In sgosline/mpnstXenoModeling: MPNST Xenograft Modeling

knitr::opts_chunk$set(echo = TRUE)
#Sys.setenv(RETICULATE_PYTHON = '/Users/bade228/opt/anaconda3/envs/r2/bin/python3')
require(remotes)

if(!require(mpnstXenoModeling)){
  remotes::install_github('sgosline/mpnstXenoModeling')
  library(mpnstXenoModeling)
}
#remotes::install_github('https://github.com/chapmandu2/IncucyteDRC')

library(data.table)
library(dplyr)
#library(repr)
library(pheatmap)
library(ggplot2)
library(drc)
library(pROC)

#syn<-mpnstXenoModeling::loadPDXData()
mpnstXenoModeling::loadSynapse()
icyteData<-mpnstXenoModeling::loadIncucyteData()%>%
  dplyr::rename(compound_name='experimentalCondition')

ggplot(data=icyteData, aes(experimental_time_point, response,colour=compound_name,shape=model_system_name)) +
geom_point()

# filter_byDrug <- function(icd, drugID) {
#   icd <- subset(icd, compound_name == drugID)
#   icd <- icd[order(icd$experimental_time_point),]
#   #icd$dosage <- log((icd$dosage)/1000000000)
#   # Assumes log(M) concentration
#   return(icd)
# }
# 
# TryFit <- function(time_response, fixed = c(NA, NA, NA, NA), names = c(NA, NA, NA, NA), nan.handle = c("LL4", "L4"), curve_type){
#     if (var(time_response$response) == 0) {
#         time_response$response[nrow(time_response)] <- time_response$response[nrow(time_response)] + 10^-10
#     }
#     #dose_response[dose_response == 0] <- 10^-10
#     nan.handle <- match.arg(nan.handle)
#     #, logDose = exp(10)
#     drug.model <- tryCatch({
#         drcmod(time_response, LL.4(fixed = fixed, names = names), curve_type)
#     }, warning = function(w) {
#         if(nan.handle == "L4"){
#             drcmod(time_response, L.4(fixed = fixed, names = names), curve_type)
#     } else {
#         drcmod(time_response, LL.4(fixed = fixed, names = names), curve_type)
#     }
#     }, error = function(e) {
#         drcmod(time_response, L.4(fixed = fixed, names = names), curve_type)
#     })
#     return(drug.model = drug.model)
# }
# drcmod <- function(time_resp, fctval, curve_type){
#   temp <- drm(formula   = response ~ experimental_time_point
#       , curveid   = model_system_name
#       , data      = time_resp
#       , fct       = fctval
#       , na.action = na.omit
#       , control   = drmc(errorm = FALSE)
#   )
#   summary(temp)
#   return(temp)
# }


# ```
# 
# ```r
# 


#res  = syn$tableQuery('select * from syn26136282')
#syn$delete(res)
#res  = syn$tableQuery('select * from syn26136282')
#synTableStore(tabname='Miccrotissue Response Valsue',tab=meta.dt,parentId='syn21984813')
#pd <- sync$build_table('Microtissue Response Values','syn21984813','doseResponse_table.csv')
#syn$store(pd)

meta.dt<-mpnstXenoModeling::process_IncucyteDrugData(icyteData, upload=FALSE)


meta.dt

Analysis of IncuCyte % confluence

icStats0 <- icyteData%>%
            mutate(maxKilling = (response/100))

minDrugs <- icStats0%>%
            mutate(hasRNASeq=(model_system_name%in%rnaSeq$Sample))%>%
            group_by(compound_name)%>%
            summarise(samples=n_distinct(model_system_name),samplesWithRNA=count(hasRNASeq==TRUE))%>%
            subset(samples>2)
#DT::datatable(minDrugs)
icStats0<-tidyr::unite(icStats0,'Drug',c(compound_name,dosage))

spi.mat<-icStats0%>%
  dplyr::select(Drug,model_system_name,maxKilling)%>%
  group_by(Drug,model_system_name)%>%
  summarise(maxKilling=max(maxKilling))%>%
  tidyr::pivot_wider(values_from=maxKilling,names_from=model_system_name,values_fill=list(maxKilling=0))%>%
  tibble::column_to_rownames('Drug')%>%
  as.matrix()
spi.mat
pheatmap(spi.mat,cellwidth = 10,cellheight=10,main='max killing across Incucyte data')
#pheatmap(spi.mat,cellwidth = 10,cellheight=10,main='max killing across Incucyte data',filename='MK_ic.png')

##let's plot the stats for each drug
icStats<-meta.dt
icStats<-tidyr::unite(icStats,'Drug',c(compound_name,dosage))

auc.mat<-icStats%>%
  dplyr::select(Drug,model_system_name,auc)%>%
  tidyr::pivot_wider(values_from=auc,names_from=model_system_name,values_fill=list(auc=0))%>%
  tibble::column_to_rownames('Drug')%>%
  as.matrix()
auc.mat
pheatmap(auc.mat, cellwidth = 10,cellheight=10,main='AUC across Incucyte data')

tgi.mat<-icStats%>%
  dplyr::select(Drug,model_system_name,ic50)%>%
  tidyr::pivot_wider(values_from=ic50,names_from=model_system_name,values_fill=list(ic50=0))%>%
  tibble::column_to_rownames('Drug')%>%
  as.matrix()

skip <- which(rowMeans(tgi.mat)%in%c(1,0))
if(length(skip)>0)
  tgi.mat <-tgi.mat[-skip,]
tgi.mat
#pheatmap(tgi.mat,cellwidth = 10,cellheight=10,main='IC50 across Incucyte data',filename='ic50_ic.png')
pheatmap(tgi.mat,cellwidth = 10,cellheight=10,main='IC50 across Incucyte data')