1 - 13 of 13 results


Employs a two-stage analysis to improve the interpretability of prediction models by preferentially using the data types upstream of gene expression. To this end, we first split the molecular data types into ‘upstream data’ (somatic mutation, copy number alteration (CNA), methylation and cancer type) and ‘downstream data’ (gene expression). This separation is based on the idea that mutation status, for example, affects the transcription of genes downstream of the pathway in which the mutation resides.

Drug response prediction

Predicts cancer drug response. Drug response prediction is based on a support vector machine (SVM) algorithm that utilizes standard recursive feature elimination (RFE) methods. It was utilized to explore the effect of a variety of alternative learning datasets on predictive accuracy leading to several unanticipated findings. This tool produces predictive scores that are plotted against observed GI50 values to graphically display the accuracy of the model employed.


Allows prediction of drug response from patient tumor gene expression data. pRRophetic can be used to predict non-clinical phenotypes, user-defined training sets and continuous and categorical phenotypes. It employs the Cancer Genome Project (CGP) cell lines as a training set. This tool’s accuracy prediction will ultimately vary from drug to drug based on the predictive power of expression under that particular set of conditions and the appropriateness of cell lines as a model of in vivo drug response.

Dr.Paso / Drug response Prediction and analysis system for oncology research

Predicts anti-cancer drug sensitivity for treatment of patients. Dr.Paso integrates network-based and statistical modelling approaches. The software figures an anti-cancer sensitivity score built on the gene expression profile of 47 genes. These genes represent hubs in a pan-cancer transcriptomic network and are implicated in several cancer-relevant biological processes. It has been evaluated on independent datasets, including in vitro validations of cell line-compound combinations.

CCLP / Cancer Cell Line Profiler

Provides differential analysis tools for high-throughput data. Cyber-T handles many types of data, from DNA and Protein microarrays, to Next Generation Sequencing (NGS), to Quantitative Mass Spectrometry (QMS). It fixes the problem of differential analysis by using the entire set of measurements to avoid instrumental or experimental biases. This software measures standard deviation (SD) via an approach derived from a Bayesian model and can achieve a regularized t-test.


A webserver developed for predicting priority and potency of an anticancer drug against a cancer cell line using its genomic features, in order to throw light in the field of personalized medicine, particularly in designing patient-specific anticancer drugs. We investigated drug profile of 24 anticancer drugs tested against a large number of cell lines in order to understand the relation between drug resistance and altered genomic features of a cancer cell line. We detected frequent mutations, high expression and high copy number variations of certain genes in both drug resistant cell lines and sensitive cell lines.

kbmtl / Kernelized Bayesian Multitask Learning

Combines kernel-based non-linear dimensionality reduction and binary classification (or regression). kbmtl main feature is the joint Bayesian formulation of projecting data points into a shared subspace and learning predictive models for all drugs in this subspace, which helps users to eliminate off-target effects and drug-specific experimental noise. This method allows users to handle missing phenotype values owing to experimental conditions and quality control reasons. It obtains statistically significantly better predictive performance on most of the drugs compared with baseline single-task algorithms that learn drug-specific models. These results show that predicting drug susceptibility against a panel of drugs simultaneously within a multitask learning framework improves overall predictive performance over single-task learning approaches.