A general calculus of fitness landscapes finds genes under selection in cancers.

Genetic variants drive the evolution of traits and diseases. We previously modeled these variants as small displacements in fitness landscapes and estimated their functional impact by differentiating the evolutionary relationship between genotype and phenotype. Conversely, here we integrate these derivatives to identify genes steering specific traits. Over cancer cohorts, integration identified 460 likely tumor-driving genes. Many have literature and experimental support but had eluded prior genomic searches for positive selection in tumors. Beyond providing cancer insights, these results introduce a general calculus of evolution to quantify the genotype-phenotype relationship and discover genes associated with complex traits and diseases.

An integrated analysis of the cancer genome atlas data discovers a hierarchical association structure across thirty three cancer types.

Cancer cells harbor molecular alterations at all levels of information processing. Genomic/epigenomic and transcriptomic alterations are inter-related between genes, within and across cancer types and may affect clinical phenotypes. Despite the abundant prior studies of integrating cancer multi-omics data, none of them organizes these associations in a hierarchical structure and validates the discoveries in extensive external data. We infer this Integrated Hierarchical Association Structure (IHAS) from the complete data of The Cancer Genome Atlas (TCGA) and compile a compendium of cancer multi-omics associations. Intriguingly, diverse alterations on genomes/epigenomes from multiple cancer types impact transcriptions of 18 Gene Groups. Half of them are further reduced to three Meta Gene Groups enriched with (1) immune and inflammatory responses, (2) embryonic development and neurogenesis, (3) cell cycle process and DNA repair. Over 80% of the clinical/molecular phenotypes reported in TCGA are aligned with the combinatorial expressions of Meta Gene Groups, Gene Groups, and other IHAS subunits. Furthermore, IHAS derived from TCGA is validated in more than 300 external datasets including multi-omics measurements and cellular responses upon drug treatments and gene perturbations in tumors, cancer cell lines, and normal tissues. To sum up, IHAS stratifies patients in terms of molecular signatures of its subunits, selects targeted genes or drugs for precision cancer therapy, and demonstrates that associations between survival times and transcriptional biomarkers may vary with cancer types. These rich information is critical for diagnosis and treatments of cancers.

Using interpretable deep learning to model cancer dependencies.

MOTIVATION: Cancer dependencies provide potential drug targets. Unfortunately, dependencies differ among cancers and even individuals. To this end, visible neural networks (VNNs) are promising due to robust performance and the interpretability required for the biomedical field. RESULTS: We design Biological visible neural network (BioVNN) using pathway knowledge to predict cancer dependencies. Despite having fewer parameters, BioVNN marginally outperforms traditional neural networks (NNs) and converges faster. BioVNN also outperforms an NN based on randomized pathways. More importantly, dependency predictions can be explained by correlating with the neuron output states of relevant pathways, which suggest dependency mechanisms. In feature importance analysis, BioVNN recapitulates known reaction partners and proposes new ones. Such robust and interpretable VNNs may facilitate the understanding of cancer dependency and the development of targeted therapies. AVAILABILITY AND IMPLEMENTATION: Code and data are available at https://github.com/LichtargeLab/BioVNN. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Discovery of disease- and drug-specific pathways through community structures of a literature network.

MOTIVATION: In light of the massive growth of the scientific literature, text mining is increasingly used to extract biological pathways. Though multiple tools explore individual connections between genes, diseases and drugs, few extensively synthesize pathways for specific diseases and drugs. RESULTS: Through community detection of a literature network, we extracted 3444 functional gene groups that represented biological pathways for specific diseases and drugs. The network linked Medical Subject Headings (MeSH) terms of genes, diseases and drugs that co-occurred in publications. The resulting communities detected highly associated genes, diseases and drugs. These significantly matched current knowledge of biological pathways and predicted future ones in time-stamped experiments. Likewise, disease- and drug-specific communities also recapitulated known pathways for those given diseases and drugs. Moreover, diseases sharing communities had high comorbidity with each other and drugs sharing communities had many common side effects, consistent with related mechanisms. Indeed, the communities robustly recovered mutual targets for drugs [area under Receiver Operating Characteristic curve (AUROC)=0.75] and shared pathogenic genes for diseases (AUROC=0.82). These data show that literature communities inform not only just known biological processes but also suggest novel disease- and drug-specific mechanisms that may guide disease gene discovery and drug repurposing. AVAILABILITY AND IMPLEMENTATION: Application tools are available at http://meteor.lichtargelab.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Multimodal network diffusion predicts future disease-gene-chemical associations.

MOTIVATION: Precision medicine is an emerging field with hopes to improve patient treatment and reduce morbidity and mortality. To these ends, computational approaches have predicted associations among genes, chemicals and diseases. Such efforts, however, were often limited to using just some available association types. This lowers prediction coverage and, since prior evidence shows that integrating heterogeneous data is likely beneficial, it may limit accuracy. Therefore, we systematically tested whether using more association types improves prediction. RESULTS: We study multimodal networks linking diseases, genes and chemicals (drugs) by applying three diffusion algorithms and varying information content. Ten-fold cross-validation shows that these networks are internally consistent, both within and across association types. Also, diffusion methods recovered missing edges, even if all the edges from an entire mode of association were removed. This suggests that information is transferable between these association types. As a realistic validation, time-stamped experiments simulated the predictions of future associations based solely on information known prior to a given date. The results show that many future published results are predictable from current associations. Moreover, in most cases, using more association types increases prediction coverage without significantly decreasing sensitivity and specificity. In case studies, literature-supported validation shows that these predictions mimic human-formulated hypotheses. Overall, this study suggests that diffusion over a more comprehensive multimodal network will generate more useful hypotheses of associations among diseases, genes and chemicals, which may guide the development of precision therapies. AVAILABILITY AND IMPLEMENTATION: Code and data are available at https://github.com/LichtargeLab/multimodal-network-diffusion. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A quantitative model for the rate-limiting process of UGA alternative assignments to stop and selenocysteine codons.

Ambiguity in genetic codes exists in cases where certain stop codons are alternatively used to encode non-canonical amino acids. In selenoprotein transcripts, the UGA codon may either represent a translation termination signal or a selenocysteine (Sec) codon. Translating UGA to Sec requires selenium and specialized Sec incorporation machinery such as the interaction between the SECIS element and SBP2 protein, but how these factors quantitatively affect alternative assignments of UGA has not been fully investigated. We developed a model simulating the UGA decoding process. Our model is based on the following assumptions: (1) charged Sec-specific tRNAs (Sec-tRNASec) and release factors compete for a UGA site, (2) Sec-tRNASec abundance is limited by the concentrations of selenium and Sec-specific tRNA (tRNASec) precursors, and (3) all synthesis reactions follow first-order kinetics. We demonstrated that this model captured two prominent characteristics observed from experimental data. First, UGA to Sec decoding increases with elevated selenium availability, but saturates under high selenium supply. Second, the efficiency of Sec incorporation is reduced with increasing selenoprotein synthesis. We measured the expressions of four selenoprotein constructs and estimated their model parameters. Their inferred Sec incorporation efficiencies did not correlate well with their SECIS-SBP2 binding affinities, suggesting the existence of additional factors determining the hierarchy of selenoprotein synthesis under selenium deficiency. This model provides a framework to systematically study the interplay of factors affecting the dual definitions of a genetic codon.

DISCOVERY OF FUNCTIONAL AND DISEASE PATHWAYS BY COMMUNITY DETECTION IN PROTEIN-PROTEIN INTERACTION NETWORKS.

Advances in cellular, molecular, and disease biology depend on the comprehensive characterization of gene interactions and pathways. Traditionally, these pathways are curated manually, limiting their efficient annotation and, potentially, reinforcing field-specific bias. Here, in order to test objective and automated identification of functionally cooperative genes, we compared a novel algorithm with three established methods to search for communities within gene interaction networks. Communities identified by the novel approach and by one of the established method overlapped significantly (q < 0.1) with control pathways. With respect to disease, these communities were biased to genes with pathogenic variants in ClinVar (p ≪ 0.01), and often genes from the same community were co-expressed, including in breast cancers. The interesting subset of novel communities, defined by poor overlap to control pathways also contained co-expressed genes, consistent with a possible functional role. This work shows that community detection based on topological features of networks suggests new, biologically meaningful groupings of genes that, in turn, point to health and disease relevant hypotheses.

Analysis of spatial-temporal gene expression patterns reveals dynamics and regionalization in developing mouse brain.

Allen Brain Atlas (ABA) provides a valuable resource of spatial/temporal gene expressions in mammalian brains. Despite rich information extracted from this database, current analyses suffer from several limitations. First, most studies are either gene-centric or region-centric, thus are inadequate to capture the superposition of multiple spatial-temporal patterns. Second, standard tools of expression analysis such as matrix factorization can capture those patterns but do not explicitly incorporate spatial dependency. To overcome those limitations, we proposed a computational method to detect recurrent patterns in the spatial-temporal gene expression data of developing mouse brains. We demonstrated that regional distinction in brain development could be revealed by localized gene expression patterns. The patterns expressed in the forebrain, medullary and pontomedullary, and basal ganglia are enriched with genes involved in forebrain development, locomotory behavior, and dopamine metabolism respectively. In addition, the timing of global gene expression patterns reflects the general trends of molecular events in mouse brain development. Furthermore, we validated functional implications of the inferred patterns by showing genes sharing similar spatial-temporal expression patterns with Lhx2 exhibited differential expression in the embryonic forebrains of Lhx2 mutant mice. These analysis outcomes confirm the utility of recurrent expression patterns in studying brain development.

Examining the condition-specific antisense transcription in S. cerevisiae and S. paradoxus.

BACKGROUND: Recent studies have demonstrated that antisense transcription is pervasive in budding yeasts and is conserved between Saccharomyces cerevisiae and S. paradoxus. While studies have examined antisense transcripts of S. cerevisiae for inverse expression in stationary phase and stress conditions, there is a lack of comprehensive analysis of the conditional specific evolutionary characteristics of antisense transcription between yeasts. Here we attempt to decipher the evolutionary relationship of antisense transcription of S. cerevisiae and S. paradoxus cultured in mid log, early stationary phase, and heat shock conditions. RESULTS: Massively parallel sequencing of sequence strand-specific cDNA library was performed from RNA isolated from S. cerevisiae and S. paradoxus cells at mid log, stationary phase and heat shock conditions. We performed this analysis using a stringent set of sense ORF transcripts and non-coding antisense transcripts that were expressed in all the three conditions, as well as in both species. We found the divergence of the condition-specific anti-sense transcription levels is higher than that in condition-specific sense transcription levels, suggesting that antisense transcription played a potential role in adapting to different conditions. Furthermore, 43% of sense-antisense pairs demonstrated inverse expression in either stationary phase or heat shock conditions relative to the mid log conditions. In addition, a large part of sense-antisense pairs (67%), which demonstrated inverse expression, were highly conserved between the two species. Our results were also concordant with known functional analyses from previous studies and with the evidence from mechanistic experiments of role of individual genes. CONCLUSIONS: By performing a genome-scale computational analysis, we have tried to evaluate the role of antisense transcription in mediating sense transcription under different environmental conditions across and in two related yeast species. Our findings suggest that antisense regulation could control expression of the corresponding sense transcript via inverse expression under a range of different circumstances.

Role of antisense RNAs in evolution of yeast regulatory complexity.

Antisense RNAs (asRNAs) are known to regulate gene expression. However, a genome-wide mechanism of asRNA regulation is unclear, and there is no good explanation why partial asRNAs are not functional. To explore its regulatory role, we investigated asRNAs using an evolutionary approach, as genome-wide experimental data are limited. We found that the percentage of genes coupling with asRNAs in Saccharomyces cerevisiae is negatively associated with regulatory complexity and evolutionary age. Nevertheless, asRNAs evolve more slowly when their sense genes are under more complex regulation. Older genes coupling with asRNAs are more likely to demonstrate inverse expression, reflecting the role of these asRNAs as repressors. Our analyses provide novel evidence, suggesting a minor contribution of asRNAs in developing regulatory complexity. Although our results support the leaky hypothesis for asRNA transcription, our evidence also suggests that partial asRNAs may have evolved as repressors. Our study deepens the understanding of asRNA regulatory evolution.

Evolution of cis-regulatory elements in yeast de novo and duplicated new genes.

BACKGROUND: New genes that originate from non-coding DNA rather than being duplicated from parent genes are called de novo genes. Their short evolution time and lack of parent genes provide a chance to study the evolution of cis-regulatory elements in the initial stage of gene emergence. Although a few reports have discussed cis-regulatory elements in new genes, knowledge of the characteristics of these elements in de novo genes is lacking. Here, we conducted a comprehensive investigation to depict the emergence and establishment of cis-regulatory elements in de novo yeast genes. RESULTS: In a genome-wide investigation, we found that the number of transcription factor binding sites (TFBSs) in de novo genes of S. cerevisiae increased rapidly and quickly became comparable to the number of TFBSs in established genes. This phenomenon might have resulted from certain characteristics of de novo genes; namely, a relatively frequent gain of TFBSs, an unexpectedly high number of preexisting TFBSs, or lower selection pressure in the promoter regions of the de novo genes. Furthermore, we identified differences in the promoter architecture between de novo genes and duplicated new genes, suggesting that distinct regulatory strategies might be employed by genes of different origin. Finally, our functional analyses of the yeast de novo genes revealed that they might be related to reproduction. CONCLUSIONS: Our observations showed that de novo genes and duplicated new genes possess mutually distinct regulatory characteristics, implying that these two types of genes might have different roles in evolution.