Defining and targeting tumor-associated macrophages in malignant mesothelioma.

Defining the ontogeny of tumor-associated macrophages (TAM) is important to develop therapeutic targets for mesothelioma. We identified two distinct macrophage populations in mouse peritoneal and pleural cavities, the monocyte-derived, small peritoneal/pleural macrophages (SPM), and the tissue-resident large peritoneal/pleural macrophages (LPM). SPM rapidly increased in tumor microenvironment after tumor challenge and contributed to the vast majority of M2-like TAM. The selective depletion of M2-like TAM by conditional deletion of the Dicer1 gene in myeloid cells (D-/-) promoted tumor rejection. Sorted SPM M2-like TAM initiated tumorigenesis in vivo and in vitro, confirming their capacity to support tumor development. The transcriptomic and single-cell RNA sequencing analysis demonstrated that both SPM and LPM contributed to the tumor microenvironment by promoting the IL-2-STAT5 signaling pathway, inflammation, and epithelial-mesenchymal transition. However, while SPM preferentially activated the KRAS and TNF-α/NFkB signaling pathways, LPM activated the IFN-γ response. The importance of LPM in the immune response was confirmed by depleting LPM with intrapleural clodronate liposomes, which abrogated the antitumoral memory immunity. SPM gene signature could be identified in pleural effusion and tumor from patients with untreated mesothelioma. Five genes, TREM2, STAB1, LAIR1, GPNMB, and MARCO, could potentially be specific therapeutic targets. Accordingly, Trem2 gene deletion led to reduced SPM M2-like TAM with compensatory increase in LPM and slower tumor growth. Overall, these experiments demonstrate that SPM M2-like TAM play a key role in mesothelioma development, while LPM more specifically contribute to the immune response. Therefore, selective targeting of monocyte-derived TAM may enhance antitumor immunity through compensatory expansion of tissue-resident TAM.

GENTANGLE: integrated computational design of gene entanglements.

SUMMARY: The design of two overlapping genes in a microbial genome is an emerging technique for adding more reliable control mechanisms in engineered organisms for increased stability. The design of functional overlapping gene pairs is a challenging procedure and computational design tools are used to improve the efficiency to deploy successful designs in genetically engineered systems. GENTANGLE (Gene Tuples ArraNGed in overLapping Elements) is a high-performance containerized pipeline for the computational design of two overlapping genes translated in different reading frames of the genome. This new software package can be used to design and test gene entanglements for microbial engineering projects using arbitrary sets of user specified gene pairs. AVAILABILITY AND IMPLEMENTATION: The GENTANGLE source code and its submodules are freely available on GitHub at https://github.com/BiosecSFA/gentangle. The DATANGLE (DATA for genTANGLE) repository contains related data and results, and is freely available on GitHub at https://github.com/BiosecSFA/datangle. The GENTANGLE container is freely available on Singularity Cloud Library at https://cloud.sylabs.io/library/khyox/gentangle/gentangle.sif. The GENTANGLE repository wiki (https://github.com/BiosecSFA/gentangle/wiki), website (https://biosecsfa.github.io/gentangle/) and user manual contain detailed instructions on how to use the different components of software and data, including examples and reproducing the results. The code is licensed under the GNU Affero General Public License version 3 (https://www.gnu.org/licenses/agpl.html). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A cross-study analysis of drug response prediction in cancer cell lines.

To enable personalized cancer treatment, machine learning models have been developed to predict drug response as a function of tumor and drug features. However, most algorithm development efforts have relied on cross-validation within a single study to assess model accuracy. While an essential first step, cross-validation within a biological data set typically provides an overly optimistic estimate of the prediction performance on independent test sets. To provide a more rigorous assessment of model generalizability between different studies, we use machine learning to analyze five publicly available cell line-based data sets: National Cancer Institute 60, ancer Therapeutics Response Portal (CTRP), Genomics of Drug Sensitivity in Cancer, Cancer Cell Line Encyclopedia and Genentech Cell Line Screening Initiative (gCSI). Based on observed experimental variability across studies, we explore estimates of prediction upper bounds. We report performance results of a variety of machine learning models, with a multitasking deep neural network achieving the best cross-study generalizability. By multiple measures, models trained on CTRP yield the most accurate predictions on the remaining testing data, and gCSI is the most predictable among the cell line data sets included in this study. With these experiments and further simulations on partial data, two lessons emerge: (1) differences in viability assays can limit model generalizability across studies and (2) drug diversity, more than tumor diversity, is crucial for raising model generalizability in preclinical screening.

Spinal Cord Stimulation for Poststroke Hemiparesis: A Scoping Review.

IMPORTANCE: Spinal cord stimulation (SCS) is a neuromodulation technique that can improve paresis in individuals with spinal cord injury. SCS is emerging as a technique that can address upper and lower limb hemiparesis. Little is understood about its effectiveness with the poststroke population. OBJECTIVE: To summarize the evidence for SCS after stroke and any changes in upper extremity and lower extremity motor function. DATA SOURCES: PubMed, Web of Science, Embase, and CINAHL. The reviewers used hand searches and reference searches of retrieved articles. There were no limitations regarding publication year. STUDY SELECTION AND DATA COLLECTION: This review followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist. The inclusion and exclusion criteria included a broad range of study characteristics. Studies were excluded if the intervention did not meet the definition of SCS intervention, used only animals or healthy participants, did not address upper or lower limb motor function, or examined neurological conditions other than stroke. FINDINGS: Fourteen articles met the criteria for this review. Seven studies found a significant improvement in motor function in groups receiving SCS. CONCLUSIONS AND RELEVANCE: Results indicate that SCS may provide an alternative means to improve motor function in the poststroke population. Plain-Language Summary: The results of this study show that spinal cord stimulation may provide an alternative way to improve motor function after stroke. Previous neuromodulation methods have targeted the impaired supraspinal circuitry after stroke. Although downregulated, spinal cord circuitry is largely intact and offers new possibilities for motor recovery.

Clustering Protein Binding Pockets and Identifying Potential Drug Interactions: A Novel Ligand-Based Featurization Method.

Protein-ligand interactions are essential to drug discovery and drug development efforts. Desirable on-target or multitarget interactions are the first step in finding an effective therapeutic, while undesirable off-target interactions are the first step in assessing safety. In this work, we introduce a novel ligand-based featurization and mapping of human protein pockets to identify closely related protein targets and to project novel drugs into a hybrid protein-ligand feature space to identify their likely protein interactions. Using structure-based template matches from PDB, protein pockets are featured by the ligands that bind to their best co-complex template matches. The simplicity and interpretability of this approach provide a granular characterization of the human proteome at the protein-pocket level instead of the traditional protein-level characterization by family, function, or pathway. We demonstrate the power of this featurization method by clustering a subset of the human proteome and evaluating the predicted cluster associations of over 7000 compounds.

A comparative genomics approach identifies contact-dependent growth inhibition as a virulence determinant.

Emerging evidence suggests the Pseudomonas aeruginosa accessory genome is enriched with uncharacterized virulence genes. Identification and characterization of such genes may reveal novel pathogenic mechanisms used by particularly virulent isolates. Here, we utilized a mouse bacteremia model to quantify the virulence of 100 individual P. aeruginosa bloodstream isolates and performed whole-genome sequencing to identify accessory genomic elements correlated with increased bacterial virulence. From this work, we identified a specific contact-dependent growth inhibition (CDI) system enriched among highly virulent P. aeruginosa isolates. CDI systems contain a large exoprotein (CdiA) with a C-terminal toxin (CT) domain that can vary between different isolates within a species. Prior work has revealed that delivery of a CdiA-CT domain upon direct cell-to-cell contact can inhibit replication of a susceptible target bacterium. Aside from mediating interbacterial competition, we observed our virulence-associated CdiA-CT domain to promote toxicity against mammalian cells in culture and lethality during mouse bacteremia. Structural and functional studies revealed this CdiA-CT domain to have in vitro tRNase activity, and mutations that abrogated this tRNAse activity in vitro also attenuated virulence. Furthermore, CdiA contributed to virulence in mice even in the absence of contact-dependent signaling. Overall, our findings indicate that this P. aeruginosa CDI system functions as both an interbacterial inhibition system and a bacterial virulence factor against a mammalian host. These findings provide an impetus for continued studies into the complex role of CDI systems in P. aeruginosa pathogenesis.

Pose Classification Using Three-Dimensional Atomic Structure-Based Neural Networks Applied to Ion Channel-Ligand Docking.

The identification of promising lead compounds showing pharmacological activities toward a biological target is essential in early stage drug discovery. With the recent increase in available small-molecule databases, virtual high-throughput screening using physics-based molecular docking has emerged as an essential tool in assisting fast and cost-efficient lead discovery and optimization. However, the best scored docking poses are often suboptimal, resulting in incorrect screening and chemical property calculation. We address the pose classification problem by leveraging data-driven machine learning approaches to identify correct docking poses from AutoDock Vina and Glide screens. To enable effective classification of docking poses, we present two convolutional neural network approaches: a three-dimensional convolutional neural network (3D-CNN) and an attention-based point cloud network (PCN) trained on the PDBbind refined set. We demonstrate the effectiveness of our proposed classifiers on multiple evaluation data sets including the standard PDBbind CASF-2016 benchmark data set and various compound libraries with structurally different protein targets including an ion channel data set extracted from Protein Data Bank (PDB) and an in-house KCa3.1 inhibitor data set. Our experiments show that excluding false positive docking poses using the proposed classifiers improves virtual high-throughput screening to identify novel molecules against each target protein compared to the initial screen based on the docking scores.

MACAW: An Accessible Tool for Molecular Embedding and Inverse Molecular Design.

The growing capabilities of synthetic biology and organic chemistry demand tools to guide syntheses toward useful molecules. Here, we present Molecular AutoenCoding Auto-Workaround (MACAW), a tool that uses a novel approach to generate molecules predicted to meet a desired property specification (e.g., a binding affinity of 50 nM or an octane number of 90). MACAW describes molecules by embedding them into a smooth multidimensional numerical space, avoiding uninformative dimensions that previous methods often introduce. The coordinates in this embedding provide a natural choice of features for accurately predicting molecular properties, which we demonstrate with examples for cetane and octane numbers, flash points, and histamine H1 receptor binding affinity. The approach is computationally efficient and well-suited to the small- and medium-size datasets commonly used in biosciences. We showcase the utility of MACAW for virtual screening by identifying molecules with high predicted binding affinity to the histamine H1 receptor and limited affinity to the muscarinic M2 receptor, which are targets of medicinal relevance. Combining these predictive capabilities with a novel generative algorithm for molecules allows us to recommend molecules with a desired property value (i.e., inverse molecular design). We demonstrate this capability by recommending molecules with predicted octane numbers of 40, 80, and 120, which is an important characteristic of biofuels. Thus, MACAW augments classical retrosynthesis tools by providing recommendations for molecules on specification.

Prehabilitation with wearables versus standard of care before major abdominal cancer surgery: a randomised controlled pilot study (trial registration: NCT04047524).

BACKGROUND: Prehabilitation aims to improve post-operative outcomes by enhancing pre-operative fitness but is labour-intensive. This pilot study aimed to assess the efficacy of a tri-modal prehabilitation programme delivered by smartwatches for improving functional fitness prior to major abdominal cancer surgery. METHODS: A single-centre, randomised controlled pilot study, in which 22 patients were randomised to: (a) a prehabilitation group (n = 11), comprising of home-based exercise, nutritional, and dietary advice delivered using a wrist-worn smartwatch connected to a smartphone application; or (b) a control group (n = 11) receiving usual care, with patients given a smartwatch as a placebo. Eligible participants had over two weeks until planned surgery. The primary outcome was pre-operative physical activity including 6-min walk test (6MWT) distance, with secondary outcomes including change in body weight and hospital anxiety and depression score (HADS). RESULTS: Recruitment was 67% of eligible patients, with groups matched for baseline characteristics. The prehabilitation group engaged in more daily minutes of moderate [25.1 min (95% CI 9.79-40.44) vs 13.1 min (95% CI 5.97-20.31), p = 0.063] and vigorous physical activity [36.1 min (95% CI 21.24-50.90) vs 17.5 min (95% CI 5.18-29.73), p = 0.022] compared to controls. They also had significantly greater improvements in 6MWT distance compared to controls [+ 85.6 m (95% CI, + 18.06 to + 153.21) vs + 13.23 m (95% CI - 6.78 to 33.23), p = 0.014]. HADS scores remained unchanged from baseline in both groups. CONCLUSION: Prehabilitation in the colorectal cancer care setting can be delivered using smartwatches and mobile applications. Furthermore, this study provides early indicative evidence that such technologies can improve functional capacity prior to surgery TRIAL REGISTRATION: NCT04047524.

Two Sides of a Coin: a Zika Virus Mutation Selected in Pregnant Rhesus Macaques Promotes Fetal Infection in Mice but at a Cost of Reduced Fitness in Nonpregnant Macaques and Diminished Transmissibility by Vectors.

Although fetal death is now understood to be a severe outcome of congenital Zika syndrome, the role of viral genetics is still unclear. We sequenced Zika virus (ZIKV) from a rhesus macaque fetus that died after inoculation and identified a single intrahost substitution, M1404I, in the ZIKV polyprotein, located in nonstructural protein 2B (NS2B). Targeted sequencing flanking position 1404 in 9 additional macaque mothers and their fetuses identified M1404I at a subconsensus frequency in the majority (5 of 9, 56%) of animals and some of their fetuses. Despite its repeated presence in pregnant macaques, M1404I has occurred rarely in humans since 2015. Since the primary ZIKV transmission cycle is human-mosquito-human, mutations in one host must be retained in the alternate host to be perpetuated. We hypothesized that ZIKV I1404 increases viral fitness in nonpregnant macaques and pregnant mice but is less efficiently transmitted by vectors, explaining its low frequency in humans during outbreaks. By examining competitive fitness relative to that of ZIKV M1404, we observed that ZIKV I1404 produced lower viremias in nonpregnant macaques and was a weaker competitor in tissues. In pregnant wild-type mice, ZIKV I1404 increased the magnitude and rate of placental infection and conferred fetal infection, in contrast to ZIKV M1404, which was not detected in fetuses. Although infection and dissemination rates were not different, Aedes aegypti mosquitoes transmitted ZIKV I1404 more poorly than ZIKV M1404. Our data highlight the complexity of arbovirus mutation-fitness dynamics and suggest that intrahost ZIKV mutations capable of augmenting fitness in pregnant vertebrates may not necessarily spread efficiently via mosquitoes during epidemics.IMPORTANCE Although Zika virus infection of pregnant women can result in congenital Zika syndrome, the factors that cause the syndrome in some but not all infected mothers are still unclear. We identified a mutation that was present in some ZIKV genomes in experimentally inoculated pregnant rhesus macaques and their fetuses. Although we did not find an association between the presence of the mutation and fetal death, we performed additional studies with ZIKV with the mutation in nonpregnant macaques, pregnant mice, and mosquitoes. We observed that the mutation increased the ability of the virus to infect mouse fetuses but decreased its capacity to produce high levels of virus in the blood of nonpregnant macaques and to be transmitted by mosquitoes. This study shows that mutations in mosquito-borne viruses like ZIKV that increase fitness in pregnant vertebrates may not spread in outbreaks when they compromise transmission via mosquitoes and fitness in nonpregnant hosts.

Machine Learning Models to Predict Inhibition of the Bile Salt Export Pump.

Cholestatic liver injury is frequently associated with drug inhibition of bile salt transporters, such as the bile salt export pump (BSEP). Reliable in silico models to predict BSEP inhibition directly from chemical structures would significantly reduce costs during drug discovery and could help avoid injury to patients. We report our development of classification and regression models for BSEP inhibition with substantially improved performance over previously published models. We assessed the performance effects of different methods of chemical featurization, data set partitioning, and class labeling and identified the methods producing models that generalized best to novel chemical entities.

Improved Protein-Ligand Binding Affinity Prediction with Structure-Based Deep Fusion Inference.

Predicting accurate protein-ligand binding affinities is an important task in drug discovery but remains a challenge even with computationally expensive biophysics-based energy scoring methods and state-of-the-art deep learning approaches. Despite the recent advances in the application of deep convolutional and graph neural network-based approaches, it remains unclear what the relative advantages of each approach are and how they compare with physics-based methodologies that have found more mainstream success in virtual screening pipelines. We present fusion models that combine features and inference from complementary representations to improve binding affinity prediction. This, to our knowledge, is the first comprehensive study that uses a common series of evaluations to directly compare the performance of three-dimensional (3D)-convolutional neural networks (3D-CNNs), spatial graph neural networks (SG-CNNs), and their fusion. We use temporal and structure-based splits to assess performance on novel protein targets. To test the practical applicability of our models, we examine their performance in cases that assume that the crystal structure is not available. In these cases, binding free energies are predicted using docking pose coordinates as the inputs to each model. In addition, we compare these deep learning approaches to predictions based on docking scores and molecular mechanic/generalized Born surface area (MM/GBSA) calculations. Our results show that the fusion models make more accurate predictions than their constituent neural network models as well as docking scoring and MM/GBSA rescoring, with the benefit of greater computational efficiency than the MM/GBSA method. Finally, we provide the code to reproduce our results and the parameter files of the trained models used in this work. The software is available as open source at https://github.com/llnl/fast. Model parameter files are available at ftp://gdo-bioinformatics.ucllnl.org/fast/pdbbind2016_model_checkpoints/.

Long-term Persistence of an Extensively Drug-Resistant Subclade of Globally Distributed Pseudomonas aeruginosa Clonal Complex 446 in an Academic Medical Center.

BACKGROUND: Antimicrobial resistance (AMR) is a major challenge in the treatment of infections caused by Pseudomonas aeruginosa. Highly drug-resistant infections are disproportionally caused by a small subset of globally distributed P. aeruginosa sequence types (STs), termed "high-risk clones." We noted that clonal complex (CC) 446 (which includes STs 298 and 446) isolates were repeatedly cultured at 1 medical center and asked whether this lineage might constitute an emerging high-risk clone. METHODS: We searched P. aeruginosa genomes from collections available from several institutions and from a public database for the presence of CC446 isolates. We determined antibacterial susceptibility using microbroth dilution and examined genome sequences to characterize the population structure of CC446 and investigate the genetic basis of AMR. RESULTS: CC446 was globally distributed over 5 continents. CC446 isolates demonstrated high rates of AMR, with 51.9% (28/54) being multidrug-resistant (MDR) and 53.6% of these (15/28) being extensively drug-resistant (XDR). Phylogenetic analysis revealed that most MDR/XDR isolates belonged to a subclade of ST298 (designated ST298*) of which 100% (21/21) were MDR and 61.9% (13/21) were XDR. XDR ST298* was identified repeatedly and consistently at a single academic medical center from 2001 through 2017. These isolates harbored a large plasmid that carries a novel antibiotic resistance integron. CONCLUSIONS: CC446 isolates are globally distributed with multiple occurrences of high AMR. The subclade ST298* is responsible for a prolonged epidemic (≥16 years) of XDR infections at an academic medical center. These findings indicate that CC446 is an emerging high-risk clone deserving further surveillance.

Binding Affinity Prediction by Pairwise Function Based on Neural Network.

We present a new approach to estimate the binding affinity from given three-dimensional poses of protein-ligand complexes. In this scheme, every protein-ligand atom pair makes an additive free-energy contribution. The sum of these pairwise contributions then gives the total binding free energy or the logarithm of the dissociation constant. The pairwise contribution is calculated by a function implemented via a neural network that takes the properties of the two atoms and their distance as input. The pairwise function is trained using a portion of the PDBbind 2018 data set. The model achieves good accuracy for affinity predictions when evaluated with PDBbind 2018 and with the CASF-2016 benchmark, comparing favorably to many scoring functions such as that of AutoDock Vina. The framework here may be extended to incorporate other factors to further improve its accuracy and power.

Predicting Small Molecule Transfer Free Energies by Combining Molecular Dynamics Simulations and Deep Learning.

Accurately predicting small molecule partitioning and hydrophobicity is critical in the drug discovery process. There are many heterogeneous chemical environments within a cell and entire human body. For example, drugs must be able to cross the hydrophobic cellular membrane to reach their intracellular targets, and hydrophobicity is an important driving force for drug-protein binding. Atomistic molecular dynamics (MD) simulations are routinely used to calculate free energies of small molecules binding to proteins, crossing lipid membranes, and solvation but are computationally expensive. Machine learning (ML) and empirical methods are also used throughout drug discovery but rely on experimental data, limiting the domain of applicability. We present atomistic MD simulations calculating 15,000 small molecule free energies of transfer from water to cyclohexane. This large data set is used to train ML models that predict the free energies of transfer. We show that a spatial graph neural network model achieves the highest accuracy, followed closely by a 3D-convolutional neural network, and shallow learning based on the chemical fingerprint is significantly less accurate. A mean absolute error of ∼4 kJ/mol compared to the MD calculations was achieved for our best ML model. We also show that including data from the MD simulation improves the predictions, tests the transferability of each model to a diverse set of molecules, and show multitask learning improves the predictions. This work provides insight into the hydrophobicity of small molecules and ML cheminformatics modeling, and our data set will be useful for designing and testing future ML cheminformatics methods.

AMPL: A Data-Driven Modeling Pipeline for Drug Discovery.

One of the key requirements for incorporating machine learning (ML) into the drug discovery process is complete traceability and reproducibility of the model building and evaluation process. With this in mind, we have developed an end-to-end modular and extensible software pipeline for building and sharing ML models that predict key pharma-relevant parameters. The ATOM Modeling PipeLine, or AMPL, extends the functionality of the open source library DeepChem and supports an array of ML and molecular featurization tools. We have benchmarked AMPL on a large collection of pharmaceutical data sets covering a wide range of parameters. Our key findings indicate that traditional molecular fingerprints underperform other feature representation methods. We also find that data set size correlates directly with prediction performance, which points to the need to expand public data sets. Uncertainty quantification can help predict model error, but correlation with error varies considerably between data sets and model types. Our findings point to the need for an extensible pipeline that can be shared to make model building more widely accessible and reproducible. This software is open source and available at: https://github.com/ATOMconsortium/AMPL.

Multidimensional Assessment of the Host Response in Mechanically Ventilated Patients with Suspected Pneumonia.

Rationale: The identification of informative elements of the host response to infection may improve the diagnosis and management of bacterial pneumonia. Objectives: To determine whether the absence of alveolar neutrophilia can exclude bacterial pneumonia in critically ill patients with suspected infection and to test whether signatures of bacterial pneumonia can be identified in the alveolar macrophage transcriptome. Methods: We determined the test characteristics of alveolar neutrophilia for the diagnosis of bacterial pneumonia in three cohorts of mechanically ventilated patients. In one cohort, we also isolated macrophages from alveolar lavage fluid and used the transcriptome to identify signatures of bacterial pneumonia. Finally, we developed a humanized mouse model of Pseudomonas aeruginosa pneumonia to determine if pathogen-specific signatures can be identified in human alveolar macrophages. Measurements and Main Results: An alveolar neutrophil percentage less than 50% had a negative predictive value of greater than 90% for bacterial pneumonia in both the retrospective (n = 851) and validation cohorts (n = 76 and n = 79). A transcriptional signature of bacterial pneumonia was present in both resident and recruited macrophages. Gene signatures from both cell types identified patients with bacterial pneumonia with test characteristics similar to alveolar neutrophilia. Conclusions: The absence of alveolar neutrophilia has a high negative predictive value for bacterial pneumonia in critically ill patients with suspected infection. Macrophages can be isolated from alveolar lavage fluid obtained during routine care and used for RNA-Seq analysis. This novel approach may facilitate a longitudinal and multidimensional assessment of the host response to bacterial pneumonia.

Diversity of Contact-Dependent Growth Inhibition Systems of Pseudomonas aeruginosa

Contact-dependent growth inhibition (CDI) systems are used in bacterial competition to hinder the growth of neighboring microbes. These systems utilize a two-partner secretion mechanism to display the CdiA exoprotein at the bacterial cell surface. CdiA forms a long filamentous stalk that facilitates binding to a target cell and delivery of a C-terminal toxin (CT) domain. This CT domain is processed and delivered into the cytoplasm of a target cell upon contact. CDI systems also encode a cognate immunity protein (CdiI) that protects siblings and resistant targeted cells from intoxication by high-affinity binding to the CT. CdiA CT domains vary among strains within a species, and many alleles encode enzymatic functions that target nucleic acids. This variation is thought to help drive diversity and adaptation within a species. CdiA diversity is well studied in Escherichia coli and several other bacteria, but little is known about the extent of this diversity in Pseudomonas aeruginosa. The purpose of this review is to highlight the variability that exists in CDI systems of P. aeruginosa. We show that this diversity is apparent even among strains isolated from a single geographical region, suggesting that CDI systems play an important role in the ecology of P. aeruginosa.

Predicting tumor cell line response to drug pairs with deep learning.

BACKGROUND: The National Cancer Institute drug pair screening effort against 60 well-characterized human tumor cell lines (NCI-60) presents an unprecedented resource for modeling combinational drug activity. RESULTS: We present a computational model for predicting cell line response to a subset of drug pairs in the NCI-ALMANAC database. Based on residual neural networks for encoding features as well as predicting tumor growth, our model explains 94% of the response variance. While our best result is achieved with a combination of molecular feature types (gene expression, microRNA and proteome), we show that most of the predictive power comes from drug descriptors. To further demonstrate value in detecting anticancer therapy, we rank the drug pairs for each cell line based on model predicted combination effect and recover 80% of the top pairs with enhanced activity. CONCLUSIONS: We present promising results in applying deep learning to predicting combinational drug response. Our feature analysis indicates screening data involving more cell lines are needed for the models to make better use of molecular features.