/PrOCTOR

A Data-Driven Approach to Predicting Successes and Failures of Clinical Trials

Primary LanguageR

An Interactive Tool for Interpretation and Testing of PrOCTOR Features and Predictions

PrOCTOR = Predicting Odds of Clinical Trial Outcomes using a Random Forest Classifier

  • Last Updated: 9/16/2016
  • Correspondence to: Katie Gayvert, kmg257 [at] cornell [dot] edu

##################

Requirements

##################

  • R - tested on version 3.2.2 (2015-08-14) -- "Fire Safety"
  • R dependecies (Feature Interpretation Tool): shiny, shinyjs, data.table, plyr, htmlwidgets, ggplot2, randomForest, grid, gridExtra, Cairo
  • Additional R dependecies (Prediction Tool): ChemmineR, ChemmineOB, rcdk, Rcpi

To Install R Dependencies:

install.packages(c("shiny", "shinyjs", "data.table", "plyr", "htmlwidgets", "ggplot2", "randomForest", "grid", "gridExtra", "Cairo","rcdk"))
source("https://bioconductor.org/biocLite.R")
biocLite("Rcpi")
biocLite("ChemmineR")
biocLite("ChemmineOB")

PrOCTOR - Prediction

##########################

To Run (in R)

##########################

source("/path/to/PrOCTOR/R/PrOCTOR.R")
PrOCTOR(SMILE,target_list)

Example (Dexamethasone):

PrOCTOR(SMILE="[H][C@@]12C[C@@H](C)[C@](O)(C(=O)CO)[C@@]1(C)C[C@H](O)[C@@]1(F)[C@@]2([H])CCC2=CC(=O)C=C[C@]12C",
        targets=c("NR3C1","NR0B1","ANXA1","NOS2"))

##########################

To Run (in R Studio)

##########################

library(shiny)
load("/path/to/PrOCTOR/shiny/PrOCTOR_prediction/initial_values.RData")
runApp("/path/to/PrOCTOR/shiny/PrOCTOR_prediction")

PrOCTOR - Feature Interpretation

##########################

To Run (in R Studio)

##########################

library(shiny)
load("/path/to/PrOCTOR/shiny/PrOCTOR_interpretation/initial_values.RData")
runApp("/path/to/PrOCTOR/shiny/PrOCTOR_interpretation")

#############

Visuals

#############

  • Feature Quantile Plot - quantile values for each feature
  • Predictions - compares each model's current prediction to distributions of failed and approved drugs in training set
  • Structure Feature Values - Current structural feature values
  • Target Feature Values - Current target-based feature values

#############

Options

#############

  • Pre-load features existing drug
  • Change individual features using manually entered values *
  • Change individual features by clicking new value on barplot *

* Change other correlated feature values along with selected feature (default set off)

#######################

Tasks In-Progress

#######################

  • Create GitHub repository
  • Make functional interface where all changes update
  • Allow user inputted structures and targets
  • Finish commenting code
  • Finish model fine-tuning