Abstract Understanding drug–drug interactions is an essential step to reduce the risk of adverse drug events before clinical co-prescription. Existing methods, commonly integrating heterogeneous data increase model performance, often suffer from a high complexity, As such, how elucidate molecular mechanisms underlying while preserving rational biological interpretability challenging task in computational modeling for discovery. In this study, we attempt investigate via associations between...

Prediction of protein subcellular localization generally involves many complex factors, and using only one or two aspects data information may not tell the true story. For this reason, some recent predictive models are deliberately designed to integrate multiple heterogeneous sources for exploiting multi-aspect feature information. Gene ontology, hereinafter referred as GO, uses a controlled vocabulary depict biological molecules gene products in terms process, molecular function cellular...

Reconstruction of host-pathogen protein interaction networks is great significance to reveal the underlying microbic pathogenesis. However, current experimentally-derived are generally small and should be augmented by computational methods for less-biased biological inference. From point view modelling, data scarcity, unavailability negative sampling three major problems reconstruction. In this work, we motivated address concerns propose a probability weighted ensemble transfer learning...

Recent years have witnessed much progress in computational modelling for protein subcellular localization. However, the existing sequence-based predictive models demonstrate moderate or unsatisfactory performance, and gene ontology (GO) based may take risk of performance overestimation novel proteins. Furthermore, many human proteins multiple locations, which renders more complicated. Up to present, there are far few researches specialized predicting localization that reside cellular...

Pathogen-host protein-protein interaction (PPI) plays an important role in revealing the underlying pathogenesis of viruses and bacteria. The need rapidly mapping proteome-wide pathogen-host interactome opens avenues for imposes burdens on computational modeling. For Salmonella typhimurium, only 62 interactions with human proteins are reported to date, modeling based such a small training data is prone yield model overfitting. In this work, we propose multi-instance transfer learning method...

Prediction of protein localization in subnuclear organelles is more challenging than general subcelluar localization. There are only three computational models for thus far, to the best our knowledge. Two were based on primary sequence only. The first model assumed homogeneous amino acid substitution pattern across all residue sites and used BLOSUM62 encode k-mer sequence. Ensemble SVM different k-mers drew final conclusion, achieving 50% overall accuracy. simplified assumption did not...

Abstract Protein-protein interaction (PPI) prediction is generally treated as a problem of binary classification wherein negative data sampling still an open to be addressed. The commonly used random prone yield less representative with considerable false negatives. Meanwhile rational constraints are seldom exerted on model selection reduce the risk positive predictions for most existing computational methods. In this work, we propose novel method based one-class SVM (support vector machine,...

Gaining knowledge about the maximum residue limits (MALs) of pesticides on fresh or processed foods is critical to process pre-harvest cultivation, post-harvest processing and storage, downstream safety surveillance food commodities. In this study, we explore available MALs 643 128 via non-negative matrix factorization (NMF) hierarchical clustering gain insights into patterns how similar exhibit profiles foods. Meanwhile, NMF predicts for untested implicitly-learnt without conducting in vivo...

Signaling pathways play important roles in the life processes of cell growth, apoptosis and organism development. At present signal transduction networks are far from complete. As an effective complement to experimental methods, computational modeling is suited rapidly reconstruct signaling at low cost. To our knowledge, existing methods seldom simultaneously exploit more than three into one predictive model for discovery novel components cross-talk between pathways.In this work, we propose...

Drug repurposing plays an important role in screening old drugs for new therapeutic efficacy. The existing methods commonly treat prediction of drug-target interaction as a problem binary classification, which large number randomly sampled pairs accounting over 50% the entire training dataset are necessarily required. Such negative examples that do not come from experimental observations inevitably decrease credibility predictions. In this study, we propose multi-label learning framework to...

Summary Identifying natural or synthetic compounds in foods and assaying their bioactivities have significantly contributed to promoting human health. In this work, we propose a machine learning framework predict 101 classes of health effects food at large scale. framework, random undersampling boosting (RUSBoost) is used as base learners tackle the problem skewed class distributions MACCSKeys similarity spectra are proposed feature engineering strategy represent chemical molecules including...

Rapid reconstruction of genome-scale protein-protein interaction (PPI) networks is instrumental in understanding the cellular processes and disease pathogenesis drug reactions. However, lack experimentally verified negative data (i.e., pairs proteins that do not interact) still a major issue needs to be properly addressed computational modeling. In this study, we take advantage very limited from Negatome infer more for We assume paralogs or orthologs two non-interacting also interact with...

Pathogen-host protein interactions are fundamental for pathogens to manipulate host signaling pathways and subvert immune defense. For most pathogens, very few or no experimental studies have been conducted investigate their cross-talks with host. In this study, we propose a computational framework validate the biological assumption that human protein-protein interaction (PPI) networks alone sufficient infer pathogen-host PPIs via pathogen functional mimicry. Pathogen mimicry assumes...

Human T-cell leukemia viruses (HTLV) tend to induce some fatal human diseases like Adult Leukemia (ATL) by targeting T lymphocytes. To indentify the protein-protein interactions (PPI) between HTLV and Homo sapiens is one of significant approaches reveal underlying mechanism infection host defence. At present, as biological experiments are labor-intensive expensive, identified part HTLV-human PPI networks rather small. Although recent years have witnessed much progress in computational...

Abstract Epstein-Barr virus (EBV) plays important roles in the origin and progression of human carcinomas, e.g. diffuse large B cell tumors, T lymphomas, etc. Discovering EBV targeted genes signaling pathways is vital to understand tumorigenesis. In this study we propose a noise-tolerant homolog knowledge transfer method reconstruct functional protein-protein interactions (PPI) networks between Homo sapiens. The training set augmented via instances noise counteracted by support vector...

Bacterial invasive infection and host immune response is fundamental to the understanding of pathogen pathogenesis discovery effective therapeutic drugs. However, there are very few experimental studies on signaling cross-talks between bacteria human date.In this work, taking M. tuberculosis H37Rv (MTB) that co-evolving with its as an example, we propose a general computational framework exploits known bacterial protein interaction networks in STRING database predict pathogen-host...

Abstract Signaling pathways play important roles in understanding the underlying mechanism of cell growth, apoptosis, organismal development and pathways-aberrant diseases. Protein-protein interaction (PPI) networks are commonly-used infrastructure to infer signaling pathways. However, PPI generally carry no information upstream/downstream relationship between interacting proteins, which retards our inferring signal flow In this work, we propose a simple feature construction method train SVM...