Systematic assessment of analytical methods for drug sensitivity prediction from cancer cell line data.

In Sock Jang, Elias Chaibub Neto, Juistin Guinney, Stephen H. Friend, Adam Margolin

Research output: Contribution to journalArticle

51 Citations (Scopus)

Abstract

Large-scale pharmacogenomic screens of cancer cell lines have emerged as an attractive pre-clinical system for identifying tumor genetic subtypes with selective sensitivity to targeted therapeutic strategies. Application of modern machine learning approaches to pharmacogenomic datasets have demonstrated the ability to infer genomic predictors of compound sensitivity. Such modeling approaches entail many analytical design choices; however, a systematic study evaluating the relative performance attributable to each design choice is not yet available. In this work, we evaluated over 110,000 different models, based on a multifactorial experimental design testing systematic combinations of modeling factors within several categories of modeling choices, including: type of algorithm, type of molecular feature data, compound being predicted, method of summarizing compound sensitivity values, and whether predictions are based on discretized or continuous response values. Our results suggest that model input data (type of molecular features and choice of compound) are the primary factors explaining model performance, followed by choice of algorithm. Our results also provide a statistically principled set of recommended modeling guidelines, including: using elastic net or ridge regression with input features from all genomic profiling platforms, most importantly, gene expression features, to predict continuous-valued sensitivity scores summarized using the area under the dose response curve, with pathway targeted compounds most likely to yield the most accurate predictors. In addition, our study provides a publicly available resource of all modeling results, an open source code base, and experimental design for researchers throughout the community to build on our results and assess novel methodologies or applications in related predictive modeling problems.

Original languageEnglish (US)
Pages (from-to)63-74
Number of pages12
JournalPacific Symposium on Biocomputing. Pacific Symposium on Biocomputing
StatePublished - 2014
Externally publishedYes

Fingerprint

Pharmacogenetics
Research Design
Cell Line
Pharmaceutical Preparations
Neoplasms
Research Personnel
Guidelines
Gene Expression
Therapeutics
Datasets
Machine Learning

ASJC Scopus subject areas

  • Medicine(all)

Cite this

Systematic assessment of analytical methods for drug sensitivity prediction from cancer cell line data. / Jang, In Sock; Neto, Elias Chaibub; Guinney, Juistin; Friend, Stephen H.; Margolin, Adam.

In: Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing, 2014, p. 63-74.

Research output: Contribution to journalArticle

@article{a9076bc53fe6456a8b338bf9a8eace02,
title = "Systematic assessment of analytical methods for drug sensitivity prediction from cancer cell line data.",
abstract = "Large-scale pharmacogenomic screens of cancer cell lines have emerged as an attractive pre-clinical system for identifying tumor genetic subtypes with selective sensitivity to targeted therapeutic strategies. Application of modern machine learning approaches to pharmacogenomic datasets have demonstrated the ability to infer genomic predictors of compound sensitivity. Such modeling approaches entail many analytical design choices; however, a systematic study evaluating the relative performance attributable to each design choice is not yet available. In this work, we evaluated over 110,000 different models, based on a multifactorial experimental design testing systematic combinations of modeling factors within several categories of modeling choices, including: type of algorithm, type of molecular feature data, compound being predicted, method of summarizing compound sensitivity values, and whether predictions are based on discretized or continuous response values. Our results suggest that model input data (type of molecular features and choice of compound) are the primary factors explaining model performance, followed by choice of algorithm. Our results also provide a statistically principled set of recommended modeling guidelines, including: using elastic net or ridge regression with input features from all genomic profiling platforms, most importantly, gene expression features, to predict continuous-valued sensitivity scores summarized using the area under the dose response curve, with pathway targeted compounds most likely to yield the most accurate predictors. In addition, our study provides a publicly available resource of all modeling results, an open source code base, and experimental design for researchers throughout the community to build on our results and assess novel methodologies or applications in related predictive modeling problems.",
author = "Jang, {In Sock} and Neto, {Elias Chaibub} and Juistin Guinney and Friend, {Stephen H.} and Adam Margolin",
year = "2014",
language = "English (US)",
pages = "63--74",
journal = "Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing",
issn = "2335-6936",

}

TY - JOUR

T1 - Systematic assessment of analytical methods for drug sensitivity prediction from cancer cell line data.

AU - Jang, In Sock

AU - Neto, Elias Chaibub

AU - Guinney, Juistin

AU - Friend, Stephen H.

AU - Margolin, Adam

PY - 2014

Y1 - 2014

N2 - Large-scale pharmacogenomic screens of cancer cell lines have emerged as an attractive pre-clinical system for identifying tumor genetic subtypes with selective sensitivity to targeted therapeutic strategies. Application of modern machine learning approaches to pharmacogenomic datasets have demonstrated the ability to infer genomic predictors of compound sensitivity. Such modeling approaches entail many analytical design choices; however, a systematic study evaluating the relative performance attributable to each design choice is not yet available. In this work, we evaluated over 110,000 different models, based on a multifactorial experimental design testing systematic combinations of modeling factors within several categories of modeling choices, including: type of algorithm, type of molecular feature data, compound being predicted, method of summarizing compound sensitivity values, and whether predictions are based on discretized or continuous response values. Our results suggest that model input data (type of molecular features and choice of compound) are the primary factors explaining model performance, followed by choice of algorithm. Our results also provide a statistically principled set of recommended modeling guidelines, including: using elastic net or ridge regression with input features from all genomic profiling platforms, most importantly, gene expression features, to predict continuous-valued sensitivity scores summarized using the area under the dose response curve, with pathway targeted compounds most likely to yield the most accurate predictors. In addition, our study provides a publicly available resource of all modeling results, an open source code base, and experimental design for researchers throughout the community to build on our results and assess novel methodologies or applications in related predictive modeling problems.

AB - Large-scale pharmacogenomic screens of cancer cell lines have emerged as an attractive pre-clinical system for identifying tumor genetic subtypes with selective sensitivity to targeted therapeutic strategies. Application of modern machine learning approaches to pharmacogenomic datasets have demonstrated the ability to infer genomic predictors of compound sensitivity. Such modeling approaches entail many analytical design choices; however, a systematic study evaluating the relative performance attributable to each design choice is not yet available. In this work, we evaluated over 110,000 different models, based on a multifactorial experimental design testing systematic combinations of modeling factors within several categories of modeling choices, including: type of algorithm, type of molecular feature data, compound being predicted, method of summarizing compound sensitivity values, and whether predictions are based on discretized or continuous response values. Our results suggest that model input data (type of molecular features and choice of compound) are the primary factors explaining model performance, followed by choice of algorithm. Our results also provide a statistically principled set of recommended modeling guidelines, including: using elastic net or ridge regression with input features from all genomic profiling platforms, most importantly, gene expression features, to predict continuous-valued sensitivity scores summarized using the area under the dose response curve, with pathway targeted compounds most likely to yield the most accurate predictors. In addition, our study provides a publicly available resource of all modeling results, an open source code base, and experimental design for researchers throughout the community to build on our results and assess novel methodologies or applications in related predictive modeling problems.

UR - http://www.scopus.com/inward/record.url?scp=84905489545&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84905489545&partnerID=8YFLogxK

M3 - Article

SP - 63

EP - 74

JO - Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing

JF - Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing

SN - 2335-6936

ER -