Heng Luo

BACKGROUND: Gene expression profiling is being widely applied in cancer research to identify biomarkers for clinical endpoint prediction. Since RNA-seq provides a powerful tool for transcriptome-based applications beyond the limitations of microarrays, we sought to systematically evaluate the performance of RNA-seq-based and microarray-based classifiers in this MAQC-III/SEQC study for clinical endpoint prediction using neuroblastoma as a model. RESULTS: We generate gene expression profiles from 498 primary neuroblastomas using both RNA-seq and 44 k microarrays. Characterization of the neuroblastoma transcriptome by RNA-seq reveals that more than 48,000 genes and 200,000 transcripts are being expressed in this malignancy. We also find that RNA-seq provides much more detailed information on specific transcript expression patterns in clinico-genetic neuroblastoma subgroups than microarrays. To systematically compare the power of RNA-seq and microarray-based models in predicting clinical endpoints, we divide the cohort randomly into training and validation sets and develop 360 predictive models on six clinical endpoints of varying predictability. Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models. CONCLUSIONS: We demonstrate that RNA-seq outperforms microarrays in determining the transcriptomic characteristics of cancer, while RNA-seq and microarray-based models perform similarly in clinical endpoint prediction. Our findings may be valuable to guide future studies on the development of gene expression-based predictive models and their implementation in clinical practice.

The rat has been used extensively as a model for evaluating chemical toxicities and for understanding drug mechanisms. However, its transcriptome across multiple organs, or developmental stages, has not yet been reported. Here we show, as part of the SEQC consortium efforts, a comprehensive rat transcriptomic BodyMap created by performing RNA-Seq on 320 samples from 11 organs of both sexes of juvenile, adolescent, adult and aged Fischer 344 rats. We catalogue the expression profiles of 40,064 genes, 65,167 transcripts, 31,909 alternatively spliced transcript variants and 2,367 non-coding genes/non-coding RNAs (ncRNAs) annotated in AceView. We find that organ-enriched, differentially expressed genes reflect the known organ-specific biological activities. A large number of transcripts show organ-specific, age-dependent or sex-specific differential expression patterns. We create a web-based, open-access rat BodyMap database of expression profiles with crosslinks to other widely used databases, anticipating that it will serve as a primary resource for biomedical research using the rat model. Gene expression is highly variable between tissues, and changes during development and with age. Here, the authors provide a comprehensive RNA-Seq analysis of the rat transcriptome, spanning eleven organs, four developmental stages and both sexes.

The identification of drug-target interactions (DTIs) is a key task in drug discovery, where drugs are chemical compounds and targets are proteins. Traditional DTI prediction methods are either time consuming (simulation-based methods) or heavily dependent on domain expertise (similarity-based and feature-based methods). In this work, we propose an end-to-end neural network model that predicts DTIs directly from low level representations. In addition to making predictions, our model provides biological interpretation using two-way attention mechanism. Instead of using simplified settings where a dataset is evaluated as a whole, we designed an evaluation dataset from BindingDB following more realistic settings where predictions of unobserved examples (proteins and drugs) have to be made. We experimentally compared our model with matrix factorization, similarity-based methods, and a previous deep learning approach. Overall, the results show that our model outperforms other approaches without requiring domain knowledge and feature engineering. In a case study, we illustrated the ability of our approach to provide biological insights to interpret the predictions.

Identifying new indications for existing drugs (drug repositioning) is an efficient way of maximizing their potential. Adverse drug reaction (ADR) is one of the leading causes of death among hospitalized patients. As both new indications and ADRs are caused by unexpected chemical-protein interactions on off-targets, it is reasonable to predict these interactions by mining the chemical-protein interactome (CPI). Making such predictions has recently been facilitated by a web server named DRAR-CPI. This server has a representative collection of drug molecules and targetable human proteins built up from our work in drug repositioning and ADR. When a user submits a molecule, the server will give the positive or negative association scores between the user's molecule and our library drugs based on their interaction profiles towards the targets. Users can thus predict the indications or ADRs of their molecule based on the association scores towards our library drugs. We have matched our predictions of drug-drug associations with those predicted via gene-expression profiles, achieving a matching rate as high as 74%. We have also successfully predicted the connections between anti-psychotics and anti-infectives, indicating the underlying relevance of anti-psychotics in the potential treatment of infections, vice versa. This server is freely available at http://cpi.bio-x.cn/drar/.

BACKGROUND: Adverse drug reactions (ADRs) are unintended and harmful reactions caused by normal uses of drugs. Predicting and preventing ADRs in the early stage of the drug development pipeline can help to enhance drug safety and reduce financial costs. METHODS: In this paper, we developed machine learning models including a deep learning framework which can simultaneously predict ADRs and identify the molecular substructures associated with those ADRs without defining the substructures a-priori. RESULTS: We evaluated the performance of our model with ten different state-of-the-art fingerprint models and found that neural fingerprints from the deep learning model outperformed all other methods in predicting ADRs. Via feature analysis on drug structures, we identified important molecular substructures that are associated with specific ADRs and assessed their associations via statistical analysis. CONCLUSIONS: The deep learning model with feature analysis, substructure identification, and statistical assessment provides a promising solution for identifying risky components within molecular structures and can potentially help to improve drug safety evaluation.