Cancer drug sensitivity estimation using modular deep Graph Neural Networks.

Computational drug sensitivity models have the potential to improve therapeutic outcomes by identifying targeted drugs components that are tailored to the transcriptomic profile of a given primary tumor. The SMILES representation of molecules that is used by state-of-the-art drug-sensitivity models is not conducive for neural networks to generalize to new drugs, in part because the distance between atoms does not generally correspond to the distance between their representation in the SMILES strings. Graph-attention networks, on the other hand, are high-capacity models that require large training-data volumes which are not available for drug-sensitivity estimation. We develop a modular drug-sensitivity graph-attentional neural network. The modular architecture allows us to separately pre-train the graph encoder and graph-attentional pooling layer on related tasks for which more data are available. We observe that this model outperforms reference models for the use cases of precision oncology and drug discovery; in particular, it is better able to predict the specific interaction between drug and cell line that is not explained by the general cytotoxicity of the drug and the overall survivability of the cell line. The complete source code is available at https://zenodo.org/doi/10.5281/zenodo.8020945. All experiments are based on the publicly available GDSC data.

Host M-CSF induced gene expression drives changes in susceptible and resistant mice-derived BMdMs upon Leishmania major infection.

Leishmaniases are a group of diseases with different clinical manifestations. Macrophage-Leishmania interactions are central to the course of the infection. The outcome of the disease depends not only on the pathogenicity and virulence of the parasite, but also on the activation state, the genetic background, and the underlying complex interaction networks operative in the host macrophages. Mouse models, with mice strains having contrasting behavior in response to parasite infection, have been very helpful in exploring the mechanisms underlying differences in disease progression. We here analyzed previously generated dynamic transcriptome data obtained from Leishmania major (L. major) infected bone marrow derived macrophages (BMdMs) from resistant and susceptible mouse. We first identified differentially expressed genes (DEGs) between the M-CSF differentiated macrophages derived from the two hosts, and found a differential basal transcriptome profile independent of Leishmania infection. These host signatures, in which 75% of the genes are directly or indirectly related to the immune system, may account for the differences in the immune response to infection between the two strains. To gain further insights into the underlying biological processes induced by L. major infection driven by the M-CSF DEGs, we mapped the time-resolved expression profiles onto a large protein-protein interaction (PPI) network and performed network propagation to identify modules of interacting proteins that agglomerate infection response signals for each strain. This analysis revealed profound differences in the resulting responses networks related to immune signaling and metabolism that were validated by qRT-PCR time series experiments leading to plausible and provable hypotheses for the differences in disease pathophysiology. In summary, we demonstrate that the host's gene expression background determines to a large degree its response to L. major infection, and that the gene expression analysis combined with network propagation is an effective approach to help identifying dynamically altered mouse strain-specific networks that hold mechanistic information about these contrasting responses to infection.

Pre-Training on In Vitro and Fine-Tuning on Patient-Derived Data Improves Deep Neural Networks for Anti-Cancer Drug-Sensitivity Prediction.

Large-scale databases that report the inhibitory capacities of many combinations of candidate drug compounds and cultivated cancer cell lines have driven the development of preclinical drug-sensitivity models based on machine learning. However, cultivated cell lines have devolved from human cancer cells over years or even decades under selective pressure in culture conditions. Moreover, models that have been trained on in vitro data cannot account for interactions with other types of cells. Drug-response data that are based on patient-derived cell cultures, xenografts, and organoids, on the other hand, are not available in the quantities that are needed to train high-capacity machine-learning models. We found that pre-training deep neural network models of drug sensitivity on in vitro drug-sensitivity databases before fine-tuning the model parameters on patient-derived data improves the models' accuracy and improves the biological plausibility of the features, compared to training only on patient-derived data. From our experiments, we can conclude that pre-trained models outperform models that have been trained on the target domains in the vast majority of cases.

Gradient tree boosting and network propagation for the identification of pan-cancer survival networks.

Cancer survival prediction is typically done with uninterpretable machine learning techniques, e.g., gradient tree boosting. Therefore, additional steps are needed to infer biological plausibility of the predictions. Here, we describe a protocol that combines pan-cancer survival prediction with XGBoost tree-ensemble learning and subsequent propagation of the learned feature weights on protein interaction networks. This protocol is based on TCGA transcriptome data of 8,024 patients from 25 cancer types but can easily be adapted to cancer patient data from other sources. For complete details on the use and execution of this protocol, please refer to Thedinga and Herwig (2022).

A gradient tree boosting and network propagation derived pan-cancer survival network of the tumor microenvironment.

Predicting cancer survival from molecular data is an important aspect of biomedical research because it allows quantifying patient risks and thus individualizing therapy. We introduce XGBoost tree ensemble learning to predict survival from transcriptome data of 8,024 patients from 25 different cancer types and show highly competitive performance with state-of-the-art methods. To further improve plausibility of the machine learning approach we conducted two additional steps. In the first step, we applied pan-cancer training and showed that it substantially improves prognosis compared with cancer subtype-specific training. In the second step, we applied network propagation and inferred a pan-cancer survival network consisting of 103 genes. This network highlights cross-cohort features and is predictive for the tumor microenvironment and immune status of the patients. Our work demonstrates that pan-cancer learning combined with network propagation generalizes over multiple cancer types and identifies biologically plausible features that can serve as biomarkers for monitoring cancer survival.

Matching anticancer compounds and tumor cell lines by neural networks with ranking loss.

Computational drug sensitivity models have the potential to improve therapeutic outcomes by identifying targeted drug components that are likely to achieve the highest efficacy for a cancer cell line at hand at a therapeutic dose. State of the art drug sensitivity models use regression techniques to predict the inhibitory concentration of a drug for a tumor cell line. This regression objective is not directly aligned with either of these principal goals of drug sensitivity models: We argue that drug sensitivity modeling should be seen as a ranking problem with an optimization criterion that quantifies a drug's inhibitory capacity for the cancer cell line at hand relative to its toxicity for healthy cells. We derive an extension to the well-established drug sensitivity regression model PaccMann that employs a ranking loss and focuses on the ratio of inhibitory concentration and therapeutic dosage range. We find that the ranking extension significantly enhances the model's capability to identify the most effective anticancer drugs for unseen tumor cell profiles based in on in-vitro data.

ClinOmicsTrailbc: a visual analytics tool for breast cancer treatment stratification.

MOTIVATION: Breast cancer is the second leading cause of cancer death among women. Tumors, even of the same histopathological subtype, exhibit a high genotypic diversity that impedes therapy stratification and that hence must be accounted for in the treatment decision-making process. RESULTS: Here, we present ClinOmicsTrailbc, a comprehensive visual analytics tool for breast cancer decision support that provides a holistic assessment of standard-of-care targeted drugs, candidates for drug repositioning and immunotherapeutic approaches. To this end, our tool analyzes and visualizes clinical markers and (epi-)genomics and transcriptomics datasets to identify and evaluate the tumor's main driver mutations, the tumor mutational burden, activity patterns of core cancer-relevant pathways, drug-specific biomarkers, the status of molecular drug targets and pharmacogenomic influences. In order to demonstrate ClinOmicsTrailbc's rich functionality, we present three case studies highlighting various ways in which ClinOmicsTrailbc can support breast cancer precision medicine. ClinOmicsTrailbc is a powerful integrated visual analytics tool for breast cancer research in general and for therapy stratification in particular, assisting oncologists to find the best possible treatment options for their breast cancer patients based on actionable, evidence-based results. AVAILABILITY AND IMPLEMENTATION: ClinOmicsTrailbc can be freely accessed at https://clinomicstrail.bioinf.uni-sb.de. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.