Screening drugs on patient biopsies from solid tumours has immense potential, but is challenging due to the small amount of available material. To address this, we present here a plug-based microfluidics platform for functional screening of drug combinations. Integrated Braille valves allow changing the plug composition on demand and enable collecting >1200 data points (56 different conditions with at least 20 replicates each) per biopsy. After deriving and validating efficient and specific drug combinations for two genetically different pancreatic cancer cell lines and xenograft mouse models, we additionally screen live cells from human solid tumours with no need for ex vivo culturing steps, and obtain highly specific sensitivity profiles. The entire workflow can be completed within 48 h at assay costs of less than US$ 150 per patient. We believe this can pave the way for rapid determination of optimal personalized cancer therapies.

Cancer cells can leverage several cell-intrinsic and -extrinsic mechanisms to escape immune system recognition. The inherent complexity of the tumor microenvironment, with its multicellular and dynamic nature, poses great challenges for the extraction of biomarkers of immune response and immunotherapy efficacy. Here, we use RNA-sequencing (RNA-seq) data combined with different sources of prior knowledge to derive system-based signatures of the tumor microenvironment, quantifying immune-cell composition and intra- and intercellular communications. We applied multi-task learning to these signatures to predict different hallmarks of immune responses and derive cancer-type-specific models based on interpretable systems biomarkers. By applying our models to independent RNA-seq data from cancer patients treated with PD-1/PD-L1 inhibitors, we demonstrated that our method to Estimate Systems Immune Response (EaSIeR) accurately predicts therapeutic outcome. We anticipate that EaSIeR will be a valuable tool to provide a holistic description of immune responses in complex and dynamic systems such as tumors using available RNA-seq data.

Genomic features are used as biomarkers of sensitivity to kinase inhibitors used widely to treat human cancer, but effective patient stratification based on these principles remains limited in impact. Insofar as kinase inhibitors interfere with signaling dynamics, and, in turn, signaling dynamics affects inhibitor responses, we investigated associations in this study between cell-specific dynamic signaling pathways and drug sensitivity. Specifically, we measured 14 phosphoproteins under 43 different perturbed conditions (combinations of 5 stimuli and 7 inhibitors) in 14 colorectal cancer cell lines, building cell line–specific dynamic logic models of underlying signaling networks. Model parameters representing pathway dynamics were used as features to predict sensitivity to a panel of 27 drugs. Specific parameters of signaling dynamics correlated strongly with drug sensitivity for 14 of the drugs, 9 of which had no genomic biomarker. Following one of these associations, we validated a drug combination predicted to overcome resistance to MEK inhibitors by coblockade of GSK3, which was not found based on associations with genomic data. These results suggest that to better understand the cancer resistance and move toward personalized medicine, it is essential to consider signaling network dynamics that cannot be inferred from static genotypes. Cancer Res; 77(12); 3364–75. ©2017 AACR.

Crowdsourcing analysis improves computational prediction of cytotoxic responses in humans. The ability to computationally predict the effects of toxic compounds on humans could help address the deficiencies of current chemical safety testing. Here, we report the results from a community-based DREAM challenge to predict toxicities of environmental compounds with potential adverse health effects for human populations. We measured the cytotoxicity of 156 compounds in 884 lymphoblastoid cell lines for which genotype and transcriptional data are available as part of the Tox21 1000 Genomes Project. The challenge participants developed algorithms to predict interindividual variability of toxic response from genomic profiles and population-level cytotoxicity data from structural attributes of the compounds. 179 submitted predictions were evaluated against an experimental data set to which participants were blinded. Individual cytotoxicity predictions were better than random, with modest correlations (Pearson's r < 0.28), consistent with complex trait genomic prediction. In contrast, predictions of population-level response to different compounds were higher (r < 0.66). The results highlight the possibility of predicting health risks associated with unknown compounds, although risk estimation accuracy remains suboptimal.

Mechanistic modeling of signaling pathways mediating patient‐specific response to therapy can help to unveil resistance mechanisms and improve therapeutic strategies. Yet, creating such models for patients, in particular for solid malignancies, is challenging. A major hurdle to build these models is the limited material available that precludes the generation of large‐scale perturbation data. Here, we present an approach that couples ex vivo high‐throughput screenings of cancer biopsies using microfluidics with logic‐based modeling to generate patient‐specific dynamic models of extrinsic and intrinsic apoptosis signaling pathways. We used the resulting models to investigate heterogeneity in pancreatic cancer patients, showing dissimilarities especially in the PI3K‐Akt pathway. Variation in model parameters reflected well the different tumor stages. Finally, we used our dynamic models to efficaciously predict new personalized combinatorial treatments. Our results suggest that our combination of microfluidic experiments and mathematical model can be a novel tool toward cancer precision medicine.

Motivation: Logic modeling is a useful tool to study signal transduction across multiple pathways. Logic models can be generated by training a network containing the prior knowledge to phospho-proteomics data. The training can be performed using stochastic optimization procedures, but these are unable to guarantee a global optima or to report the complete family of feasible models. This, however, is essential to provide precise insight in the mechanisms underlaying signal transduction and generate reliable predictions. Results: We propose the use of Answer Set Programming to explore exhaustively the space of feasible logic models. Toward this end, we have developed caspo, an open-source Python package that provides a powerful platform to learn and characterize logic models by leveraging the rich modeling language and solving technologies of Answer Set Programming. We illustrate the usefulness of caspo by revisiting a model of pro-growth and inflammatory pathways in liver cells. We show that, if experimental error is taken into account, there are thousands (11 700) of models compatible with the data. Despite the large number, we can extract structural features from the models, such as links that are always (or never) present or modules that appear in a mutual exclusive fashion. To further characterize this family of models, we investigate the input–output behavior of the models. We find 91 behaviors across the 11 700 models and we suggest new experiments to discriminate among them. Our results underscore the importance of characterizing in a global and exhaustive manner the family of feasible models, with important implications for experimental design. Availability:caspo is freely available for download (license GPLv3) and as a web service at http://caspo.genouest.org/. Supplementary information: Supplementary materials are available at Bioinformatics online. Contact:anne.siegel@irisa.fr or saezrodriguez@ebi.ac.uk