Community assessment to advance computational prediction of cancer drug combinations in a pharmacogenomic screen

Menden, Michael P.; Wang, Dennis; Mason, Mike J.; Szalai, Bence; Bulusu, Krishna C.; Guan, Yuanfang; Yu, Thomas; Kang, Jaewoo; Jeon, Minji; Wolfinger, Russ; Nguyen, Tin; Zaslavskiy, Mikhail; Jang, In Sock; Ghazoui, Zara; Ahsen, Mehmet Eren; Vogel, Robert; Neto, Elias Chaibub; Norman, Thea; Tang, Jing; Garnett, Mathew J.; Veroli, Giovanni Y. Di; Fawell, Stephen E.; Stolovitzky, Gustavo; Guinney, Justin; Dry, Jonathan R.; Sáez-Rodríguez, Julio

doi:10.1038/s41467-019-09799-2

Cited by 282 publications

(271 citation statements)

References 53 publications

Supporting

Mentioning

263

Contrasting

Order By: Relevance

“…A recurrent observation across DREAM challenges is that an ensemble of individual predictions performs usually better and is more robust than any individual method [16, 17]. This phenomenon, common also in other contexts, is denoted as the wisdom-of-the-crowds (WOC) [15].…”

Section: Resultsmentioning

confidence: 99%

Predicting cellular position in the Drosophila embryo from Single-Cell Transcriptomics data

Tanevski

Nguyen

Truong

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

Single-cell RNA-seq (scRNAseq) technologies are rapidly evolving and a growing number of datasets are now available. While very informative, in standard scRNAseq experiments the spatial organization of the cells in the organism or tissue of origin is lost. Conversely, spatial RNA-seq technologies designed to keep the localization of the cells have limited throughput and gene coverage. Mapping scRNAseq to data of genes with spatial information can thus increase coverage while providing spatial location. However, methods to perform such a mapping are still in their infancy and have not been benchmarked in an unbiased manner. To bridge the gap, we organized the DREAM Single-Cell Transcriptomics challenge to evaluate methods for reconstructing the spatial arrangement of single cells from single-cell RNA sequencing data. The challenge focused on the spatial reconstruction of cells from the Drosophila embryo from single-cell transcriptomic and, leveraging as gold standard, in situ hybridization data of a set of selected driver genes from the Berkeley Drosophila Transcription Network Project reference atlas. The 34 participating teams used an array of different algorithms for gene selection and location prediction. We devised a novel scoring and cross-validation scheme to evaluate the robustness of the best performing algorithms. Participants were able to correctly and robustly localize rare subpopulations of cells, accurately mapping both spatially co-localized and scattered groups of cells. The selection of predictor genes was essential for accurately locating the cells in the embryo. Among the most frequently selected set of genes we measured a relatively high expression entropy, high spatial clustering and the presence of prominent developmental genes such as gap and pair-ruled genes and tissue defining markers. IntroductionThe recent technological advances in single-cell sequencing technologies have revolutionized 1 the biological sciences. In particular single-cell RNA sequencing (scRNAseq) methods allow 2 transcriptome profiling in a highly parallel manner, resulting in the quantification of thousands of 3 genes across thousands of cells of the same tissue. However, with a few exceptions [1, 2, 3, 4, 5] 4 current high-throughput scRNAseq methods share the drawback of losing the information relative 5 to the spatial arrangement of the cells in the tissue during the cell dissociation step. 6 One way of regaining spatial information computationally is to appropriately combine the single-7 cell RNA dataset at hand with a reference database, or atlas, containing spatial expression patterns 8 for several genes across the tissue. This approach was pursued in a few studies [6, 7, 8, 9, 10]. 9 Achim et al identified the location of 139 cells using 72 reference genes with spatial information 10 from whole mount in situ hybridization (WMISH) of a marine annelid and Satija et al developed 11 the Seurat algorithm to predict position of 851 zebrafish cells based on their scRNAseq data and 12 spatial information from in s...

show abstract

Section: Resultsmentioning

confidence: 99%

Predicting cellular position in the Drosophila embryo from Single-Cell Transcriptomics data

Tanevski

Nguyen

Truong

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…We obtained global insight into the principles and functions of interactions from analysis of the landscape, and validated new hypotheses about combinatorial cytokine effects. Developing learning models to predict synergistic combinations of treatments based on the individual effects is an active area of research [18][19][20][21] . Despite an apparent methodological similarity, the motivation and goals of our framework are fundamentally different from these studies.…”

Section: Discussionmentioning

confidence: 99%

Deciphering the combinatorial interaction landscape

Cappuccio

Jensen

Hartmann

et al. 2019

Preprint

View full text Add to dashboard Cite

From cellular activation to drug combinations, the control of biological systems involves multiple stimuli that can elicit complex nonlinear interactions. To elucidate the functions and logic of stimulus interactions, we developed SAIL (Synergistic/Antagonistic Interaction Learner). SAIL uses a machine learning classifier trained to categorize interactions across a complete taxonomy of possible combinatorial effects. The strategy resolves the most informative interactions, and helps infer their functions and regulatory mechanisms. SAIL-predicted interaction mechanisms controlling key immune functions were experimentally validated. SAIL can integrate results from multiple datasets to derive general properties of how cells respond to multiple stimuli. Using public immunological datasets, we assembled a fine-grained landscape of~30000 interactions. Analysis of the landscape shows the context-dependent functions of individual modulators, and reveals a probabilistic algebra that links the separate and combined stimulus effects. SAIL is available through a user friendly interface to resolve the effect of stimulus and drug combinations .

show abstract

“…To this end, several algorithms have been proposed to predict synergistic drug pairs which utilize a diverse set of features such as chemical structure, biological networks interactions (e.g., drug-protein, protein -disease, etc.) and omics data [6,7,8,9,10,11,12,13,14,15]. While feature engineering and a systems approach come with the promise of performance increase, features other than the chemical structure are not always available, and biological networks such as protein interaction networks are incomplete.…”

Section: Introductionmentioning

confidence: 99%

MatchMaker: A Deep Learning Framework for Drug Synergy Prediction

Kuru

Taştan

Çiçek

2020

Preprint

View full text Add to dashboard Cite

Drug combination therapies have been a viable strategy for the treatment of complex diseases such as cancer due to increased efficacy and reduced side effects. However, experimentally validating all possible combinations for synergistic interaction even with highthroughout screens is intractable due to vast combinatorial search space. Computational techniques can reduce the number of combinations to be evaluated experimentally by prioritizing promising candidates. We present MatchMaker that predicts drug synergy scores using drug chemical structure information and gene expression profiles of cell lines in a deep learning framework. For the first time, our model utilizes the largest known drug combination dataset to date, DrugComb. We compare the performance of MatchMaker with the state-of-the-art models and observe up to ∼ 20% correlation and ∼ 40% mean squared error (MSE) improvements over the next best method. We investigate the cell types and drug pairs that are relatively harder to predict and present novel candidate pairs. MatchMaker is built and available at https://github.com/tastanlab/matchmaker

show abstract

Community assessment to advance computational prediction of cancer drug combinations in a pharmacogenomic screen

Cited by 282 publications

References 53 publications

Predicting cellular position in the Drosophila embryo from Single-Cell Transcriptomics data

Predicting cellular position in the Drosophila embryo from Single-Cell Transcriptomics data

Deciphering the combinatorial interaction landscape

MatchMaker: A Deep Learning Framework for Drug Synergy Prediction

Contact Info

Product

Resources

About