Hybrid Gene Origination Creates Human-Virus Chimeric Proteins during Infection.

RNA viruses are a major human health threat. The life cycles of many highly pathogenic RNA viruses like influenza A virus (IAV) and Lassa virus depends on host mRNA, because viral polymerases cleave 5'-m7G-capped host transcripts to prime viral mRNA synthesis ("cap-snatching"). We hypothesized that start codons within cap-snatched host transcripts could generate chimeric human-viral mRNAs with coding potential. We report the existence of this mechanism of gene origination, which we named "start-snatching." Depending on the reading frame, start-snatching allows the translation of host and viral "untranslated regions" (UTRs) to create N-terminally extended viral proteins or entirely novel polypeptides by genetic overprinting. We show that both types of chimeric proteins are made in IAV-infected cells, generate T cell responses, and contribute to virulence. Our results indicate that during infection with IAV, and likely a multitude of other human, animal and plant viruses, a host-dependent mechanism allows the genesis of hybrid genes.

Transcription Elongation Can Affect Genome 3D Structure.

How transcription affects genome 3D organization is not well understood. We found that during influenza A (IAV) infection, rampant transcription rapidly reorganizes host cell chromatin interactions. These changes occur at the ends of highly transcribed genes, where global inhibition of transcription termination by IAV NS1 protein causes readthrough transcription for hundreds of kilobases. In these readthrough regions, elongating RNA polymerase II disrupts chromatin interactions by inducing cohesin displacement from CTCF sites, leading to locus decompaction. Readthrough transcription into heterochromatin regions switches them from the inert (B) to the permissive (A) chromatin compartment and enables transcription factor binding. Data from non-viral transcription stimuli show that transcription similarly affects cohesin-mediated chromatin contacts within gene bodies. Conversely, inhibition of transcription elongation allows cohesin to accumulate at previously transcribed intragenic CTCF sites and to mediate chromatin looping and compaction. Our data indicate that transcription elongation by RNA polymerase II remodels genome 3D architecture.

Chemical Proteomics Identifies Druggable Vulnerabilities in a Genetically Defined Cancer.

The transcription factor NRF2 is a master regulator of the cellular antioxidant response, and it is often genetically activated in non-small-cell lung cancers (NSCLCs) by, for instance, mutations in the negative regulator KEAP1. While direct pharmacological inhibition of NRF2 has proven challenging, its aberrant activation rewires biochemical networks in cancer cells that may create special vulnerabilities. Here, we use chemical proteomics to map druggable proteins that are selectively expressed in KEAP1-mutant NSCLC cells. Principal among these is NR0B1, an atypical orphan nuclear receptor that we show engages in a multimeric protein complex to regulate the transcriptional output of KEAP1-mutant NSCLC cells. We further identify small molecules that covalently target a conserved cysteine within the NR0B1 protein interaction domain, and we demonstrate that these compounds disrupt NR0B1 complexes and impair the anchorage-independent growth of KEAP1-mutant cancer cells. Our findings designate NR0B1 as a druggable transcriptional regulator that supports NRF2-dependent lung cancers.

Position-dependent function of human sequence-specific transcription factors.

Patterns of transcriptional activity are encoded in our genome through regulatory elements such as promoters or enhancers that, paradoxically, contain similar assortments of sequence-specific transcription factor (TF) binding sites1-3. Knowledge of how these sequence motifs encode multiple, often overlapping, gene expression programs is central to understanding gene regulation and how mutations in non-coding DNA manifest in disease4,5. Here, by studying gene regulation from the perspective of individual transcription start sites (TSSs), using natural genetic variation, perturbation of endogenous TF protein levels and massively parallel analysis of natural and synthetic regulatory elements, we show that the effect of TF binding on transcription initiation is position dependent. Analysing TF-binding-site occurrences relative to the TSS, we identified several motifs with highly preferential positioning. We show that these patterns are a combination of a TF's distinct functional profiles-many TFs, including canonical activators such as NRF1, NFY and Sp1, activate or repress transcription initiation depending on their precise position relative to the TSS. As such, TFs and their spacing collectively guide the site and frequency of transcription initiation. More broadly, these findings reveal how similar assortments of TF binding sites can generate distinct gene regulatory outcomes depending on their spatial configuration and how DNA sequence polymorphisms may contribute to transcription variation and disease and underscore a critical role for TSS data in decoding the regulatory information of our genome.

Functional landscape of SARS-CoV-2 cellular restriction.

A deficient interferon (IFN) response to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection has been implicated as a determinant of severe coronavirus disease 2019 (COVID-19). To identify the molecular effectors that govern IFN control of SARS-CoV-2 infection, we conducted a large-scale gain-of-function analysis that evaluated the impact of human IFN-stimulated genes (ISGs) on viral replication. A limited subset of ISGs were found to control viral infection, including endosomal factors inhibiting viral entry, RNA binding proteins suppressing viral RNA synthesis, and a highly enriched cluster of endoplasmic reticulum (ER)/Golgi-resident ISGs inhibiting viral assembly/egress. These included broad-acting antiviral ISGs and eight ISGs that specifically inhibited SARS-CoV-2 and SARS-CoV-1 replication. Among the broad-acting ISGs was BST2/tetherin, which impeded viral release and is antagonized by SARS-CoV-2 Orf7a protein. Overall, these data illuminate a set of ISGs that underlie innate immune control of SARS-CoV-2/SARS-CoV-1 infection, which will facilitate the understanding of host determinants that impact disease severity and offer potential therapeutic strategies for COVID-19.

Breakthrough infections by SARS-CoV-2 variants boost cross-reactive hybrid immune responses in mRNA-vaccinated Golden Syrian hamsters.

Hybrid immunity (vaccination + natural infection) to SARS-CoV-2 provides superior protection to re-infection. We performed immune profiling studies during breakthrough infections in mRNA-vaccinated hamsters to evaluate hybrid immunity induction. The mRNA vaccine, BNT162b2, was dosed to induce binding antibody titers against ancestral spike, but inefficient serum virus neutralization of ancestral SARS-CoV-2 or variants of concern (VoCs). Vaccination reduced morbidity and controlled lung virus titers for ancestral virus and Alpha but allowed breakthrough infections in Beta, Delta and Mu-challenged hamsters. Vaccination primed for T cell responses that were boosted by infection. Infection back-boosted neutralizing antibody responses against ancestral virus and VoCs. Hybrid immunity resulted in more cross-reactive sera, reflected by smaller antigenic cartography distances. Transcriptomics post-infection reflects both vaccination status and disease course and suggests a role for interstitial macrophages in vaccine-mediated protection. Therefore, protection by vaccination, even in the absence of high titers of neutralizing antibodies in the serum, correlates with recall of broadly reactive B- and T-cell responses.

Discovery of SARS-CoV-2 antiviral drugs through large-scale compound repurposing.

The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in 2019 has triggered an ongoing global pandemic of the severe pneumonia-like disease coronavirus disease 2019 (COVID-19)1. The development of a vaccine is likely to take at least 12-18 months, and the typical timeline for approval of a new antiviral therapeutic agent can exceed 10 years. Thus, repurposing of known drugs could substantially accelerate the deployment of new therapies for COVID-19. Here we profiled a library of drugs encompassing approximately 12,000 clinical-stage or Food and Drug Administration (FDA)-approved small molecules to identify candidate therapeutic drugs for COVID-19. We report the identification of 100 molecules that inhibit viral replication of SARS-CoV-2, including 21 drugs that exhibit dose-response relationships. Of these, thirteen were found to harbour effective concentrations commensurate with probable achievable therapeutic doses in patients, including the PIKfyve kinase inhibitor apilimod2-4 and the cysteine protease inhibitors MDL-28170, Z LVG CHN2, VBY-825 and ONO 5334. Notably, MDL-28170, ONO 5334 and apilimod were found to antagonize viral replication in human pneumocyte-like cells derived from induced pluripotent stem cells, and apilimod also demonstrated antiviral efficacy in a primary human lung explant model. Since most of the molecules identified in this study have already advanced into the clinic, their known pharmacological and human safety profiles will enable accelerated preclinical and clinical evaluation of these drugs for the treatment of COVID-19.

Triclosan administration to humanized UDP-glucuronosyltransferase 1 neonatal mice induces UGT1A1 through a dependence on PPARα and ATF4.

Triclosan (TCS) is an antimicrobial toxicant found in a myriad of consumer products and has been detected in human tissues, including breastmilk. We have evaluated the impact of lactational TCS on UDP-glucuronosyltransferase 1A1 (UGT1A1) expression and bilirubin metabolism in humanized UGT1 (hUGT1) neonatal mice. In hUGT1 mice, expression of the hepatic UGT1A1 gene is developmentally delayed resulting in elevated total serum bilirubin (TSB) levels. We found that newborn hUGT1 mice breastfed or orally treated with TCS presented lower TSB levels along with induction of hepatic UGT1A1. Lactational and oral treatment by gavage with TCS leads to the activation of hepatic nuclear receptors constitutive androstane receptor (CAR), peroxisome proliferator-activated receptor alpha (PPARα), and stress sensor, activating transcription factor 4 (ATF4). When CAR-deficient hUGT1 mice (hUGT1/Car-/-) were treated with TCS, TSB levels were reduced with a robust induction of hepatic UGT1A1, leaving us to conclude that CAR is not tied to UGT1A1 induction. Alternatively, when PPARα-deficient hUGT1 mice (hUGT1/Pparα-/-) were treated with TCS, hepatic UGT1A1 was not induced. Additionally, we had previously demonstrated that TCS is a potent inducer of ATF4, a transcriptional factor linked to the integrated stress response. When ATF4 was deleted in liver of hUGT1 mice (hUGT1/Atf4ΔHep) and these mice treated with TCS, we observed superinduction of hepatic UGT1A1. Oxidative stress genes in livers of hUGT1/Atf4ΔHep treated with TCS were increased, suggesting that ATF4 protects liver from excessive oxidative stress. The increase oxidative stress may be associated with superinduction of UGT1A1. The expression of ATF4 in neonatal hUGT1 hepatic tissue may play a role in the developmental repression of UGT1A1.

MHC class II proteins mediate cross-species entry of bat influenza viruses.

Zoonotic influenza A viruses of avian origin can cause severe disease in individuals, or even global pandemics, and thus pose a threat to human populations. Waterfowl and shorebirds are believed to be the reservoir for all influenza A viruses, but this has recently been challenged by the identification of novel influenza A viruses in bats1,2. The major bat influenza A virus envelope glycoprotein, haemagglutinin, does not bind the canonical influenza A virus receptor, sialic acid or any other glycan1,3,4, despite its high sequence and structural homology with conventional haemagglutinins. This functionally uncharacterized plasticity of the bat influenza A virus haemagglutinin means the tropism and zoonotic potential of these viruses has not been fully determined. Here we show, using transcriptomic profiling of susceptible versus non-susceptible cells in combination with genome-wide CRISPR-Cas9 screening, that the major histocompatibility complex class II (MHC-II) human leukocyte antigen DR isotype (HLA-DR) is an essential entry determinant for bat influenza A viruses. Genetic ablation of the HLA-DR α-chain rendered cells resistant to infection by bat influenza A virus, whereas ectopic expression of the HLA-DR complex in non-susceptible cells conferred susceptibility. Expression of MHC-II from different bat species, pigs, mice or chickens also conferred susceptibility to infection. Notably, the infection of mice with bat influenza A virus resulted in robust virus replication in the upper respiratory tract, whereas mice deficient for MHC-II were resistant. Collectively, our data identify MHC-II as a crucial entry mediator for bat influenza A viruses in multiple species, which permits a broad vertebrate tropism.

Oral arsenic administration to humanizedUDP-glucuronosyltransferase1 neonatal mice induces UGT1A1 through a dependence on Nrf2 and PXR.

Inorganic arsenic (iAs) is an environmental toxicant that can lead to severe health consequences, which can be exacerbated if exposure occurs early in development. Here, we evaluated the impact of oral iAs treatment on UDP-glucuronosyltransferase 1A1 (UGT1A1) expression and bilirubin metabolism in humanized UGT1 (hUGT1) mice. We found that oral administration of iAs to neonatal hUGT1 mice that display severe neonatal hyperbilirubinemia leads to induction of intestinal UGT1A1 and a reduction in total serum bilirubin values. Oral iAs administration accelerates neonatal intestinal maturation, an event that is directly associated with UGT1A1 induction. As a reactive oxygen species producer, oral iAs treatment activated the Keap-Nrf2 pathway in the intestinal tract and liver. When Nrf2-deficient hUGT1 mice (hUGT1/Nrf2-/-) were treated with iAs, it was shown that activated Nrf2 contributed significantly toward intestinal maturation and UGT1A1 induction. However, hepatic UGT1A1 was not induced upon iAs exposure. We previously demonstrated that the nuclear receptor PXR represses liver UGT1A1 in neonatal hUGT1 mice. When PXR was deleted in hUGT1 mice (hUGT1/Pxr-/-), derepression of UGT1A1 was evident in both liver and intestinal tissue in neonates. Furthermore, when neonatal hUGT1/Pxr-/- mice were treated with iAs, UGT1A1 was superinduced in both tissues, confirming PXR release derepressed key regulatory elements on the gene that could be activated by iAs exposure. With iAs capable of generating reactive oxygen species in both liver and intestinal tissue, we conclude that PXR deficiency in neonatal hUGT1/Pxr-/- mice allows greater access of activated transcriptional modifiers such as Nrf2 leading to superinduction of UGT1A1.

Identification and dynamic quantification of regulatory elements using total RNA.

The spatial and temporal regulation of transcription initiation is pivotal for controlling gene expression. Here, we introduce capped-small RNA-seq (csRNA-seq), which uses total RNA as starting material to detect transcription start sites (TSSs) of both stable and unstable RNAs at single-nucleotide resolution. csRNA-seq is highly sensitive to acute changes in transcription and identifies an order of magnitude more regulated transcripts than does RNA-seq. Interrogating tissues from species across the eukaryotic kingdoms identified unstable transcripts resembling enhancer RNAs, pri-miRNAs, antisense transcripts, and promoter upstream transcripts in multicellular animals, plants, and fungi spanning 1.6 billion years of evolution. Integration of epigenomic data from these organisms revealed that histone H3 trimethylation (H3K4me3) was largely confined to TSSs of stable transcripts, whereas H3K27ac marked nucleosomes downstream from all active TSSs, suggesting an ancient role for posttranslational histone modifications in transcription. Our findings show that total RNA is sufficient to identify transcribed regulatory elements and capture the dynamics of initiated stable and unstable transcripts at single-nucleotide resolution in eukaryotes.

A Genome-Wide CRISPR-Cas9 Screen Reveals the Requirement of Host Cell Sulfation for Schmallenberg Virus Infection.

Schmallenberg virus (SBV) is an insect-transmitted orthobunyavirus that can cause abortions and congenital malformations in the offspring of ruminants. Even though the two viral surface glycoproteins Gn and Gc are involved in host cell entry, the specific cellular receptors of SBV are currently unknown. Using genome-wide CRISPR-Cas9 forward screening, we identified 3'-phosphoadenosine 5'-phosphosulfate (PAPS) transporter 1 (PAPST1) as an essential factor for SBV infection. PAPST1 is a sulfotransferase involved in heparan sulfate proteoglycan synthesis encoded by the solute carrier family 35 member B2 gene (SLC35B2). SBV cell surface attachment and entry were largely reduced upon the knockout of SLC35B2, whereas the reconstitution of SLC35B2 in these cells fully restored their susceptibility to SBV infection. Furthermore, treatment of cells with heparinase diminished infection with SBV, confirming that heparan sulfate plays an important role in cell attachment and entry, although to various degrees, heparan sulfate was also found to be important to initiate infection by two other bunyaviruses, La Crosse virus and Rift Valley fever virus. Thus, PAPST1-triggered synthesis of cell surface heparan sulfate is required for the efficient replication of SBV and other bunyaviruses.IMPORTANCE SBV is a newly emerging orthobunyavirus (family Peribunyaviridae) that has spread rapidly across Europe since 2011, resulting in substantial economic losses in livestock farming. In this study, we performed unbiased genome-wide CRISPR-Cas9 screening and identified PAPST1, a sulfotransferase encoded by SLC35B2, as a host entry factor for SBV. Consistent with its role in the synthesis of heparan sulfate, we show that this activity is required for efficient infection by SBV. A comparable dependency on heparan sulfate was also observed for La Crosse virus and Rift Valley fever virus, highlighting the importance of heparan sulfate for host cell infection by bunyaviruses. Thus, the present work provides crucial insights into virus-host interactions of important animal and human pathogens.

Viral Determinants in H5N1 Influenza A Virus Enable Productive Infection of HeLa Cells.

Influenza A virus (IAV) is a human respiratory pathogen that causes yearly global epidemics, as well as sporadic pandemics due to human adaptation of pathogenic strains. Efficient replication of IAV in different species is, in part, dictated by its ability to exploit the genetic environment of the host cell. To investigate IAV tropism in human cells, we evaluated the replication of IAV strains in a diverse subset of epithelial cell lines. HeLa cells were refractory to the growth of human H1N1 and H3N2 viruses and low-pathogenic avian influenza (LPAI) viruses. Interestingly, a human isolate of the highly pathogenic avian influenza (HPAI) H5N1 virus successfully propagated in HeLa cells to levels comparable to those in a human lung cell line. Heterokaryon cells generated by fusion of HeLa and permissive cells supported H1N1 virus growth, suggesting the absence of a host factor(s) required for the replication of H1N1, but not H5N1, viruses in HeLa cells. The absence of this factor(s) was mapped to reduced nuclear import, replication, and translation, as well as deficient viral budding. Using reassortant H1N1:H5N1 viruses, we found that the combined introduction of nucleoprotein (NP) and hemagglutinin (HA) from an H5N1 virus was necessary and sufficient to enable H1N1 virus growth. Overall, this study suggests that the absence of one or more cellular factors in HeLa cells results in abortive replication of H1N1, H3N2, and LPAI viruses, which can be circumvented upon the introduction of H5N1 virus NP and HA. Further understanding of the molecular basis of this restriction will provide important insights into the virus-host interactions that underlie IAV pathogenesis and tropism.IMPORTANCE Many zoonotic avian influenza A viruses have successfully crossed the species barrier and caused mild to life-threatening disease in humans. While human-to-human transmission is limited, there is a risk that these zoonotic viruses may acquire adaptive mutations enabling them to propagate efficiently and cause devastating human pandemics. Therefore, it is important to identify viral determinants that provide these viruses with a replicative advantage in human cells. Here, we tested the growth of influenza A virus in a subset of human cell lines and found that abortive replication of H1N1 viruses in HeLa cells can be circumvented upon the introduction of H5N1 virus HA and NP. Overall, this work leverages the genetic diversity of multiple human cell lines to highlight viral determinants that could contribute to H5N1 virus pathogenesis and tropism.

SHMT2 and the BRCC36/BRISC deubiquitinase regulate HIV-1 Tat K63-ubiquitylation and destruction by autophagy.

HIV-1 Tat is a key regulator of viral transcription, however little is known about the mechanisms that control its turnover in T cells. Here we use a novel proteomics technique, called DiffPOP, to identify the molecular target of JIB-04, a small molecule compound that potently and selectively blocks HIV-1 Tat expression, transactivation, and virus replication in T cell lines. Mass-spectrometry analysis of whole-cell extracts from 2D10 Jurkat T cells revealed that JIB-04 targets Serine Hydroxymethyltransferase 2 (SHMT2), a regulator of glycine biosynthesis and an adaptor for the BRCC36 K63Ub-specific deubiquitinase in the BRISC complex. Importantly, knockdown of SHMT1,2 or BRCC36, or exposure of cells to JIB-04, strongly increased Tat K63Ub-dependent destruction via autophagy. Moreover, point mutation of multiple lysines in Tat, or knockdown of BRCC36 or SHMT1,2, was sufficient to prevent destruction of Tat by JIB-04. We conclude that HIV-1 Tat levels are regulated through K63Ub-selective autophagy mediated through SHMT1,2 and the BRCC36 deubiquitinase.

Massively Parallel Sequencing of Peritoneal and Splenic B Cell Repertoires Highlights Unique Properties of B-1 Cell Antibodies.

B-1 cells are a unique subset of B cells that are positively selected for expressing autoreactive BCRs. We isolated RNA from peritoneal (B-1a, B-1b, B-2) and splenic (B-1a, marginal zone, follicular) B cells from C57BL/6 mice and used 5'-RACE to amplify the IgH V region using massively parallel sequencing. By analyzing 379,000 functional transcripts, we demonstrate that B-1a cells use a distinct and restricted repertoire. All B-1 cell subsets, especially peritoneal B-1a cells, had a high proportion of sequences without N additions, suggesting predominantly prenatal development. Their transcripts differed markedly and uniquely contained VH11 and VH12 genes, which were rearranged only with a restricted selection of D and J genes, unlike other V genes. Compared to peritoneal B-1a, the peritoneal B-1b repertoire was larger, had little overlap with B-1a, and most sequences contained N additions. Similarly, the splenic B-1a repertoire differed from peritoneal B-1a sequences, having more unique sequences and more frequent N additions, suggesting influx of B-1a cells into the spleen from nonperitoneal sites. Two CDR3s, previously described as Abs to bromelain-treated RBCs, comprised 43% of peritoneal B-1a sequences. We show that a single-chain variable fragment designed after the most prevalent B-1a sequence bound oxidation-specific epitopes such as the phosphocholine of oxidized phospholipids. In summary, we provide the IgH V region library of six murine B cell subsets, including, to our knowledge for the first time, a comparison between B-1a and B-1b cells, and we highlight qualities of B-1 cell Abs that indicate unique selection processes.

ApoC-III ASO promotes tissue LPL activity in the absence of apoE-mediated TRL clearance.

Hypertriglyceridemia results from accumulation of triglyceride (TG)-rich lipoproteins (TRLs) in the circulation and is associated with increased CVD risk. ApoC-III is an apolipoprotein on TRLs and a prominent negative regulator of TG catabolism. We recently established that in vivo apoC-III predominantly inhibits LDL receptor-mediated and LDL receptor-related protein 1-mediated hepatic TRL clearance and that apoC-III-enriched TRLs are preferentially cleared by syndecan-1 (SDC1). In this study, we determined the impact of apoE, a common ligand for all three receptors, on apoC-III metabolism using apoC-III antisense oligonucleotide (ASO) treatment in mice lacking apoE and functional SDC1 (Apoe-/-Ndst1f/fAlb-Cre+). ApoC-III ASO treatment significantly reduced plasma TG levels in Apoe-/-Ndst1f/fAlb-Cre+ mice without reducing hepatic VLDL production or improving hepatic TRL clearance. Further analysis revealed that apoC-III ASO treatment lowered plasma TGs in Apoe-/-Ndst1f/fAlb-Cre+ mice, which was associated with increased LPL activity in white adipose tissue in the fed state. Finally, clinical data confirmed that ASO-mediated lowering of APOC-III via volanesorsen can reduce plasma TG levels independent of the APOE isoform genotype. Our data indicate that apoE determines the metabolic impact of apoC-III as we establish that apoE is essential to mediate inhibition of TRL clearance by apoC-III and that, in the absence of functional apoE, apoC-III inhibits tissue LPL activity.

Deep sequencing of protease inhibitor resistant HIV patient isolates reveals patterns of correlated mutations in Gag and protease.

While the role of drug resistance mutations in HIV protease has been studied comprehensively, mutations in its substrate, Gag, have not been extensively cataloged. Using deep sequencing, we analyzed a unique collection of longitudinal viral samples from 93 patients who have been treated with therapies containing protease inhibitors (PIs). Due to the high sequence coverage within each sample, the frequencies of mutations at individual positions were calculated with high precision. We used this information to characterize the variability in the Gag polyprotein and its effects on PI-therapy outcomes. To examine covariation of mutations between two different sites using deep sequencing data, we developed an approach to estimate the tight bounds on the two-site bivariate probabilities in each viral sample, and the mutual information between pairs of positions based on all the bounds. Utilizing the new methodology we found that mutations in the matrix and p6 proteins contribute to continued therapy failure and have a major role in the network of strongly correlated mutations in the Gag polyprotein, as well as between Gag and protease. Although covariation is not direct evidence of structural propensities, we found the strongest correlations between residues on capsid and matrix of the same Gag protein were often due to structural proximity. This suggests that some of the strongest inter-protein Gag correlations are the result of structural proximity. Moreover, the strong covariation between residues in matrix and capsid at the N-terminus with p1 and p6 at the C-terminus is consistent with residue-residue contacts between these proteins at some point in the viral life cycle.

CoVaMa: Co-Variation Mapper for disequilibrium analysis of mutant loci in viral populations using next-generation sequence data.

Next-Generation Sequencing (NGS) has transformed our understanding of the dynamics and diversity of virus populations for human pathogens and model systems alike. Due to the sensitivity and depth of coverage in NGS, it is possible to measure the frequency of mutations that may be present even at vanishingly low frequencies within the viral population. Here, we describe a simple bioinformatic pipeline called CoVaMa (Co-Variation Mapper) scripted in Python that detects correlated patterns of mutations in a viral sample. Our algorithm takes NGS alignment data and populates large matrices of contingency tables that correspond to every possible pairwise interaction of nucleotides in the viral genome or amino acids in the chosen open reading frame. These tables are then analysed using classical linkage disequilibrium to detect and report evidence of epistasis. We test our analysis with simulated data and then apply the approach to find epistatically linked loci in Flock House Virus genomic RNA grown under controlled cell culture conditions. We also reanalyze NGS data from a large cohort of HIV infected patients and find correlated amino acid substitution events in the protease gene that have arisen in response to anti-viral therapy. This both confirms previous findings and suggests new pairs of interactions within HIV protease. The script is publically available at http://sourceforge.net/projects/covama.

Drug target prediction through deep learning functional representation of gene signatures.

Many machine learning applications in bioinformatics currently rely on matching gene identities when analyzing input gene signatures and fail to take advantage of preexisting knowledge about gene functions. To further enable comparative analysis of OMICS datasets, including target deconvolution and mechanism of action studies, we develop an approach that represents gene signatures projected onto their biological functions, instead of their identities, similar to how the word2vec technique works in natural language processing. We develop the Functional Representation of Gene Signatures (FRoGS) approach by training a deep learning model and demonstrate that its application to the Broad Institute's L1000 datasets results in more effective compound-target predictions than models based on gene identities alone. By integrating additional pharmacological activity data sources, FRoGS significantly increases the number of high-quality compound-target predictions relative to existing approaches, many of which are supported by in silico and/or experimental evidence. These results underscore the general utility of FRoGS in machine learning-based bioinformatics applications. Prediction networks pre-equipped with the knowledge of gene functions may help uncover new relationships among gene signatures acquired by large-scale OMICs studies on compounds, cell types, disease models, and patient cohorts.