Limits...
Predicting Anticancer Drug Responses Using a Dual-Layer Integrated Cell Line-Drug Network Model.

Zhang N, Wang H, Fang Y, Wang J, Zheng X, Liu XS - PLoS Comput. Biol. (2015)

Bottom Line: When using the dual-layer model integrating both CSN and DSN, our predicted response reached a 0.6 Pearson correlation coefficient with observed responses for most drugs, which is significantly better than the previous results using the elastic net model.We have also applied the dual-layer cell line-drug integrated network model to fill in the missing drug response values in the CGP dataset.Even though the dual-layer integrated cell line-drug network model does not specifically model mutation information, it correctly predicted that BRAF mutant cell lines would be more sensitive than BRAF wild-type cell lines to three MEK1/2 inhibitors tested.

View Article: PubMed Central - PubMed

Affiliation: Department of Mathematics, Shanghai Normal University, Shanghai, China.

ABSTRACT
The ability to predict the response of a cancer patient to a therapeutic agent is a major goal in modern oncology that should ultimately lead to personalized treatment. Existing approaches to predicting drug sensitivity rely primarily on profiling of cancer cell line panels that have been treated with different drugs and selecting genomic or functional genomic features to regress or classify the drug response. Here, we propose a dual-layer integrated cell line-drug network model, which uses both cell line similarity network (CSN) data and drug similarity network (DSN) data to predict the drug response of a given cell line using a weighted model. Using the Cancer Cell Line Encyclopedia (CCLE) and Cancer Genome Project (CGP) studies as benchmark datasets, our single-layer model with CSN or DSN and only a single parameter achieved a prediction performance comparable to the previously generated elastic net model. When using the dual-layer model integrating both CSN and DSN, our predicted response reached a 0.6 Pearson correlation coefficient with observed responses for most drugs, which is significantly better than the previous results using the elastic net model. We have also applied the dual-layer cell line-drug integrated network model to fill in the missing drug response values in the CGP dataset. Even though the dual-layer integrated cell line-drug network model does not specifically model mutation information, it correctly predicted that BRAF mutant cell lines would be more sensitive than BRAF wild-type cell lines to three MEK1/2 inhibitors tested.

No MeSH data available.


Related in: MedlinePlus

Model assumption.(A, B) Box plots showing cell lines with similar gene expression profiles responding similarly to the same drugs. The X-axis indicates the Pearson correlation coefficients between all possible cell line pairs based on expression profiles. The Y-axis shows the correlations of their drug response vectors as measured by activity area in CCLE (A) and CGP (B). (C, D) Box plots showing that drugs with similar 1-D and 2-D structural features based on PaDEL exhibiting similar effects on cell lines in the CCLE (C) and CGP (D) datasets. The X-axis represents the drug similarity categories, and the Y-axis shows the correlations of drug responses across all cell lines. Statistical differences between two groups were measured by the t-test.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4587957&req=5

pcbi.1004498.g001: Model assumption.(A, B) Box plots showing cell lines with similar gene expression profiles responding similarly to the same drugs. The X-axis indicates the Pearson correlation coefficients between all possible cell line pairs based on expression profiles. The Y-axis shows the correlations of their drug response vectors as measured by activity area in CCLE (A) and CGP (B). (C, D) Box plots showing that drugs with similar 1-D and 2-D structural features based on PaDEL exhibiting similar effects on cell lines in the CCLE (C) and CGP (D) datasets. The X-axis represents the drug similarity categories, and the Y-axis shows the correlations of drug responses across all cell lines. Statistical differences between two groups were measured by the t-test.

Mentions: We built the dual-layer integrated cell line-drug network model using large pharmacogenomics datasets from the CCLE [13] and CGP [14] studies. Experimentally determined drug responses, also referred to as drug sensitivities, which were measured as activity area and IC50 in both studies. Notably, a higher value of activity area or lower value of IC50 indicates a better sensitivity of a cell line to a given drug. We first took activity area as drug response measurement. The CCLE study, for example, contains expression profiles of 491 cancer cell lines, as well as their response to 24 drugs. Since different drugs have different baseline values and ranges, we normalized the drug response data so that different drugs have the same baseline value and range across all cell lines. We calculated the Pearson correlation of gene expression profiles and the Pearson correlation of drug responses by activity area for each cell line pair. Drug sensitivity correlations were significantly higher for cell lines with more similar gene expression profiles (Fig 1A). The CGP dataset contains response data and expression profiles for 653 cell lines treated with 139 drugs. In agreement with the CCLE observations, CGP cell lines with higher gene expression similarity show higher drug response correlations for all the drugs tested (Fig 1B). These results suggest that cell lines with similar gene expression profiles exhibit similar drug responses.


Predicting Anticancer Drug Responses Using a Dual-Layer Integrated Cell Line-Drug Network Model.

Zhang N, Wang H, Fang Y, Wang J, Zheng X, Liu XS - PLoS Comput. Biol. (2015)

Model assumption.(A, B) Box plots showing cell lines with similar gene expression profiles responding similarly to the same drugs. The X-axis indicates the Pearson correlation coefficients between all possible cell line pairs based on expression profiles. The Y-axis shows the correlations of their drug response vectors as measured by activity area in CCLE (A) and CGP (B). (C, D) Box plots showing that drugs with similar 1-D and 2-D structural features based on PaDEL exhibiting similar effects on cell lines in the CCLE (C) and CGP (D) datasets. The X-axis represents the drug similarity categories, and the Y-axis shows the correlations of drug responses across all cell lines. Statistical differences between two groups were measured by the t-test.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4587957&req=5

pcbi.1004498.g001: Model assumption.(A, B) Box plots showing cell lines with similar gene expression profiles responding similarly to the same drugs. The X-axis indicates the Pearson correlation coefficients between all possible cell line pairs based on expression profiles. The Y-axis shows the correlations of their drug response vectors as measured by activity area in CCLE (A) and CGP (B). (C, D) Box plots showing that drugs with similar 1-D and 2-D structural features based on PaDEL exhibiting similar effects on cell lines in the CCLE (C) and CGP (D) datasets. The X-axis represents the drug similarity categories, and the Y-axis shows the correlations of drug responses across all cell lines. Statistical differences between two groups were measured by the t-test.
Mentions: We built the dual-layer integrated cell line-drug network model using large pharmacogenomics datasets from the CCLE [13] and CGP [14] studies. Experimentally determined drug responses, also referred to as drug sensitivities, which were measured as activity area and IC50 in both studies. Notably, a higher value of activity area or lower value of IC50 indicates a better sensitivity of a cell line to a given drug. We first took activity area as drug response measurement. The CCLE study, for example, contains expression profiles of 491 cancer cell lines, as well as their response to 24 drugs. Since different drugs have different baseline values and ranges, we normalized the drug response data so that different drugs have the same baseline value and range across all cell lines. We calculated the Pearson correlation of gene expression profiles and the Pearson correlation of drug responses by activity area for each cell line pair. Drug sensitivity correlations were significantly higher for cell lines with more similar gene expression profiles (Fig 1A). The CGP dataset contains response data and expression profiles for 653 cell lines treated with 139 drugs. In agreement with the CCLE observations, CGP cell lines with higher gene expression similarity show higher drug response correlations for all the drugs tested (Fig 1B). These results suggest that cell lines with similar gene expression profiles exhibit similar drug responses.

Bottom Line: When using the dual-layer model integrating both CSN and DSN, our predicted response reached a 0.6 Pearson correlation coefficient with observed responses for most drugs, which is significantly better than the previous results using the elastic net model.We have also applied the dual-layer cell line-drug integrated network model to fill in the missing drug response values in the CGP dataset.Even though the dual-layer integrated cell line-drug network model does not specifically model mutation information, it correctly predicted that BRAF mutant cell lines would be more sensitive than BRAF wild-type cell lines to three MEK1/2 inhibitors tested.

View Article: PubMed Central - PubMed

Affiliation: Department of Mathematics, Shanghai Normal University, Shanghai, China.

ABSTRACT
The ability to predict the response of a cancer patient to a therapeutic agent is a major goal in modern oncology that should ultimately lead to personalized treatment. Existing approaches to predicting drug sensitivity rely primarily on profiling of cancer cell line panels that have been treated with different drugs and selecting genomic or functional genomic features to regress or classify the drug response. Here, we propose a dual-layer integrated cell line-drug network model, which uses both cell line similarity network (CSN) data and drug similarity network (DSN) data to predict the drug response of a given cell line using a weighted model. Using the Cancer Cell Line Encyclopedia (CCLE) and Cancer Genome Project (CGP) studies as benchmark datasets, our single-layer model with CSN or DSN and only a single parameter achieved a prediction performance comparable to the previously generated elastic net model. When using the dual-layer model integrating both CSN and DSN, our predicted response reached a 0.6 Pearson correlation coefficient with observed responses for most drugs, which is significantly better than the previous results using the elastic net model. We have also applied the dual-layer cell line-drug integrated network model to fill in the missing drug response values in the CGP dataset. Even though the dual-layer integrated cell line-drug network model does not specifically model mutation information, it correctly predicted that BRAF mutant cell lines would be more sensitive than BRAF wild-type cell lines to three MEK1/2 inhibitors tested.

No MeSH data available.


Related in: MedlinePlus