Biology subjects

Rodriguez, J. A.

Publications and source records attributed to Rodriguez, J. A..

At least 19 recordsLinked to original sources

bioRxiv2026

Computationally hybridized pathogenic mammarenavirus receptor binding domains reveal cryptic variants that elicit cross-neutralizing immune responses

The pathogenic mammarenaviruses, Machupo (MACV) and Junin (JUNV), are under evolutionary pressure to leverage human transferrin receptor 1 (hTfR1) for cellular entry while evading host immune responses during zoonosis. We now structurally and functionally investigate cryptic, computationally hybridized JUNV-MACV GP1 sequence variants. We then evaluate the ability of those variants to facilitate internalization of pseudotyped virus-like particles (PVs) into human cells and their recognition by neutralizing antibodies, including plasma from JUNV convalescent patients. We further compare the cryoEM structures of hTfR1-bound MACV GP1 to those of two functional hTfR1-bound MACV-JUNV hybrid GP1 variants that enable robust PV internalization and are also recognized by cross-neutralizing antibodies. Immunization with these variants demonstrates they and other hybrid GP1 sequences can elicit cross-reactive, neutralizing antibodies, supporting a model in which sequence adaptation within GP1 balances receptor recognition with immune evasion. This may inform the rational design of broadly neutralizing GP1-targeted antiviral therapies.

biophysics↗

bioRxiv2026

Cryo-EM provides insight into how the Staphylococcus aureus IsdH receptor removes hemin from the hemoglobin:haptoglobin complex

Staphylococcus aureus extracts hemin from human hemoglobin (Hb) to overcome host-imposed iron limitation. How it recovers Hb-bound hemin from the hemoglobin:haptoglobin (Hb:Hp) complex, the major circulating form of Hb outside red blood cells, remains unclear. Here we use cryo-electron microscopy, biophysical measurements, and solution kinetics to define how the S. aureus IsdH surface receptor extracts hemin from Hb:Hp. A 3.1 [A] cryo-EM structure of Hb:Hp bound by full-length IsdH reveals that its N-terminal NEAT domain (N1) anchors it to Hb, whereas its downstream N2N3 extraction unit engages {beta}Hb to remove its hemin. The receptor engages Hb:Hp differently than isolated Hb, because N-linked glycans on haptoglobin bias the extraction unit toward {beta}Hb, sterically occluding its access to Hb while still permitting engagement by N1. Kinetic assays show that IsdH actively accelerates hemin release from Hb:Hp. Three-dimensional variability analysis indicates that this likely occurs via a dynamic interface in which receptor motions reposition the extraction unit relative to {beta}Hb, collectively supporting a model in which IsdH transiently perturbs the F-helix to promote hemin extraction. Alignment of that model with a previously determined CD163:Hb:Hp structure shows how IsdH may disrupt Hb:Hp recognition by macrophage and monocyte CD163 receptors, helping to explain how it may hinder clearance of Hb:Hp from circulation. In aggregate, these results help define the structural basis for hemin extraction from Hb:Hp and how IsdH may subvert receptor-mediated clearance of the Hb:Hp complex. SignificanceStaphylococcus aureus scavenges hemin from host hemoglobin to proliferate, yet most extracellular hemoglobin is sequestered in hemoglobin:haptoglobin (Hb:Hp) complexes, which are rapidly cleared from circulation. To clarify how bacteria access hemin in this context, we now show that the IsdH surface receptor is structurally adapted to extract hemin from Hb:Hp. Our results indicate that IsdH uses distinct NEAT domains to anchor to Hb and selectively extract hemin from {beta}Hb. This {beta}Hb selectivity is shaped by haptoglobin N-glycans and enhances microbial access to iron. These findings provide a mechanistic framework for targeting heme acquisition as an anti-virulence strategy.

microbiology↗

bioRxiv2026

Visual exoproteomics of Clostridium thermocellum during anaerobic biomass-degradation identifies functional spirosomes

Visual proteomics enables the study of low-abundance proteins and identification of unknown complexes from heterogeneous samples by complementing high-resolution cryogenic electron microscopy (cryoEM) with external inputs on protein identity such as mass spectrometry. Using this approach, we interrogated the exoproteome of the anaerobic cellulose-degrading bacterium Clostridium thermocellum as it carried out biomass degradation. Mass spectrometry indicated a broad exoproteome composition, including cellulose degrading machinery CelA and CipA. A focus on large exoproteome assemblies revealed abundant protein filaments and pleomorphic vesicular structures. Analysis of the most abundant protein filaments yielded an [~]4 {square} resolution native structure that, aided by mass spectrometry, de novo modeling, and structural searching, was found to be the aldehyde-alcohol dehydrogenase (AdhE) spirosome. AdhE contained both NAD+ and Fe in their expected binding sites and biochemical and structural analyses of enriched spirosome preparations indicated they were functional. Altered NADH solution concentrations triggered conformational changes in the exoproteomic spirosomes, and the constituent AdhE remained capable of ethanol production. Although the basis for functional extracellular spirosome accumulation in live anaerobic C. thermocellum cultures remains unclear, their abundance in crude exoproteomes suggests their presence could influence biomass fueled C. thermocellum growth.

biochemistry↗

bioRxiv2025

Leveraging bioorthogonal conjugation for alpha synuclein fibril surveillance

Alpha synuclein (-syn) amyloid fibrils are associated with various neurodegenerative diseases. To better understand the molecular and cellular basis for -syn fibril persistence and spread, we implemented a fluorophore labeling strategy to surveil pre-formed -syn fibrils in solution and in cells. We leveraged amber codon mediated incorporation of a tetrazine-based artificial amino acid (TetV2.0) to install a cyclooctene-conjugated Janeliaflour, JF549, at four sites on human -syn: residues 4, 60, 96 and 136. Fast coupling occurred under mild buffer conditions and in the presence of the disease-associated cofactor and cytotoxic lipid, psychosine. Labeled fibrils retained their polymorphic features, seeded the growth of new fibrils in vitro, and induced the seeding of positive puncta in -syn FRET biosensor HEK293T cells. This allowed simultaneous tracking of exogenous and endogenous -syn aggregates in biosensor cells, and their localization within the cells. In doing so, our approach facilitates more detailed mechanistic investigation of -syn aggregates.

biochemistry↗

bioRxiv2025

Evolutionary genomics reveals plant origins of acetic acid bacteria in fermented food

Humans have historically relied on acetic acid bacteria (AAB) for food fermentation, yet their origins must trace back to free-living species outside of human environments. In nature, plants, fruit flies, and social insects host AAB. However, the evolutionary transitions of AAB from symbiotic hosts to fermented foods remain ambiguous. Here, we conduct a comprehensive phylogenomic analysis of 570 publicly available AAB genomes. We find that the [~]170My evolutionary history of this group is concordant with the rise of angiosperms, corbiculate bees, and the consequential accelerated availability of environmental carbohydrates. Unlike other ferment-associated microbes, ferment-associated AAB have exclusively evolved from clades inhabiting flowers and fruits, but not insect hosts. Genomic features are similar in plant- and ferment-associated AAB, yet markers of early adaptation to ferments are also present. Conversely, social insect-associated AAB have reduced genome sizes, which may have limited their functional capability to disperse into ferments. Plant- and ferment-associated AAB coincide in the ability to metabolise diverse plant carbohydrates, though both have adapted to produce habitat-specific carbohydrate-active enzymes. In contrast, metabolic capacity is reduced in social insect-associated AAB. By tracing the phylogenomics of this clade, we understand how evolution forged AAB capable of performing metabolic work for humans, shaping the history and potential futures of fermentation.

evolutionary biology↗

bioRxiv2025

Structural basis for L-isoaspartyl-containing protein recognition by the PCMTD1 cullin-RING E3 ubiquitin ligase

A major type of spontaneous protein damage that accumulates with age is the formation of kinked polypeptide chains with L-isoaspartyl residues. Mitigating this damage is necessary for maintaining proteome stability and prolonging organismal survival. While repair through methylation by PCMT1 has been previously shown to suppress L-isoaspartyl accumulation, we provide an additional mechanism for L-isoaspartyl maintenance through PCMTD1, a cullin-RING ligase (CRL). We combined cryo-EM, native mass spectrometry, and biochemical assays to provide insight on how the assembly and architecture of PCMTD1 in the context of a CRL complex fulfils this alternative mechanism. We show that the PCMTD1 CRL complex specifically binds L-isoaspartyl residues when bound to AdoMet. This work provides evidence for a growing class of E3 ubiquitin ligases that recognize spontaneous covalent modifications as potential substrates for ubiquitylation and subsequent proteasomal degradation. eTOC BlurbLimiting the accrual of L-isoaspartyl damaged proteins is essential during aging. While this is thought to be mediated solely by the repair activity of the protein, PCMT1, Pang et al. now demonstrate that a related protein, PCMTD1, functions as a cullin-RING ligase to selectively target L-isoaspartyl-damaged substrates for potential regulation by the ubiquitylation-proteosomal system. HighlightsO_LIAtomic cryo-EM structure of CRL5-PCMTD1 determined C_LIO_LIArchitecture of PCMTD1 when complexed as a CRL supports ubiquitylation activity C_LIO_LIPCMTD1 recognizes L-isoaspartyl residues as a recruitment motif for potential CRL activity C_LIO_LIRecognition of L-isoaspartyl residues is dependent on cofactor engagement C_LI

biochemistry↗

bioRxiv2025

The effect of different milk pretreatment methods on microbiome community development during Herrgards cheese production and ripening.

One of the biggest challenges for dairy producers is the substantial variability in final product properties caused by changes in the production environment. In cheese production, this variation is influenced by several factors, particularly the milk base and its pretreatment, which shape the microbiome throughout the process and ultimately affect the cheeses organoleptic characteristics. To examine the impact of three different pre-treatments for pasteurised milk-- microfiltration, protein fortification, and only pasteurisation (control)-- on microbiome dynamics, we generated metagenome sequencing data from 14 cheese production steps across these three production trials at a Danish dairy factory. We constructed three metagenomic co-assemblies, identifying nine high-quality metagenome-assembled genomes (MAGs). Our analysis revealed that a specific strain of Lactococcus lactis dominates the process, while other minor bacterial species persist at very low abundances (<1%), contributing non-negligibly to product properties. Notably, Clostridium tyrobutyricum, a known dairy spoilage bacterium, was present at low levels in pasteurised-only and protein-fortified milk trials but was nearly absent in microfiltered milk. To enhance our analyses, we implemented KHILL, a novel k-mer-based method applied directly to raw sequencing reads, which facilitates metagenomic co-assembly and enables early detection of unwanted microorganisms. Our findings provide industrial dairy producers with a comprehensive view of microbial dynamics during cheese production, offering insights to improve process consistency and product quality.

microbiology↗

bioRxiv2025

Combining MicroED and native mass spectrometry for structural discovery of enzyme-biosynthetic inhibitor complexes

With the goal of accelerating the discovery of small molecule-protein complexes, we leverage fast, low-dose, event based electron counting microcrystal electron diffraction (MicroED) data collection and native mass spectrometry. This approach resolves structures of the epoxide-based cysteine protease inhibitor, and natural product, E-64, and its biosynthetic analogs bound to the model cysteine protease, papain. The combined structural power of MicroED and the analytical capabilities of native mass spectrometry (ED-MS) allows assignment of papain structures bound to E-64-like ligands with data obtained from crystal slurries soaked with mixtures of known inhibitors, and crude biosynthetic reactions. ED-MS further discriminates the highest-affinity ligand soaked into microcrystals from a broad inhibitor cocktail, and identifies multiple similarly high-affinity ligands soaked into microcrystals simultaneously. This extends to libraries of printed ligands dispensed directly onto TEM grids and later soaked with papain microcrystal slurries. ED-MS identifies papain binding to its preferred natural products, by showing that two analogues of E-64 outcompete others in binding to papain crystals, and by detecting papain bound to E-64 and an analogue from crude biosynthetic reactions, without purification. This illustrates the utility of ED-MS for natural product ligand discovery and for structure-based screening of small molecule binders to macromolecular targets.

biophysics↗

bioRxiv2024

Persistent genomic erosion in whooping cranes despite demographic recovery

Integrating in-situ (wild) and ex-situ (captive) conservation efforts can mitigate genetic diversity loss and help prevent extinction of endangered wild populations. The whooping crane (Grus americana) experienced severe population declines in the 18th century, culminating into a collapse to 16 individuals in 1941. Legal protections and conservation actions have since increased the population to approximately 840 individuals, yet the impact on genomic diversity remains unclear. We analysed the temporal dynamics of genomic erosion by sequencing a high-quality genome reference, and re-sequencing 16 historical and 37 modern genomes, including wild individuals and four generations of captive-bred individuals. Genomic demographic reconstructions reveal a steady decline, accelerating over the past 300 years with the European settlement of North America. Temporal genomic analyses show that despite demographic recovery, the species has lost 70% of its genetic diversity and has increased their inbreeding. Although the modern population bottleneck reduced the ancestral genetic load, modern populations possess more realized load than masked load, possibly resulting in a chronic loss of fitness. Integrating pedigree and genomic data, we underscore the role of breeding management in reducing recent inbreeding. Yet ongoing heterozygosity loss, load accumulation, and background inbreeding argues against the species downlisting from their current Endangered status on the IUCN Red List and the Endangered Species Act. The presence of private genetic variation in wild and captive populations suggests that wild-captive crosses could enhance genetic diversity and reduce the realized load. Our findings emphasize the role of genomics in informing conservation management and policy.

evolutionary biology↗

bioRxiv2024

Biosynthetic characterization and combinatorial biocatalysis of the cysteine protease inhibitor E-64

E-64 is an irreversible and selective cysteine protease inhibitor prominently used in chemical biology and drug discovery. In this work, we uncovered and characterized the NRPS-independent pathway responsible for biosynthesis of E-64, which is widely conserved in fungi. Heterologous reconstitution and biochemical assays show the pathway starts with epoxidation of fumaric acid to the warhead (2S,3S)-trans-epoxysuccinic acid with an -ketoglutarate (KG)/Fe(II)-dependent oxygenase, followed by successive condensation with an O_SCPLOWLC_SCPLOW-amino acid by an ATP-grasp enzyme, and with an amine by the first characterized amide bond synthetase from fungi. Both amide bond-forming enzymes displayed significant biocatalytic potential, including scalability, stereoselectivity towards the warhead and broader substrate scopes in forming the amide bonds. Combinatorial biocatalysis with the two amide-bond forming enzymes generated a library of cysteine protease inhibitors and led to more potent analogs towards cathepsin B. In addition, preparative synthesis of clinically relevant cysteine protease inhibitors was accomplished from a single reaction mixture. Our work highlights the importance of biosynthetic investigation for enzyme discovery and the potential of amide bond-forming enzymes as biocatalysts for a library synthesis of small molecules.

biochemistry↗

bioRxiv2023

Exosomal microRNA signature from plasma-derived extracellular vesicles in gastric cancer

BackgroundGastric cancer is a heterogeneous pathology that represents the fifth most frequent malignancy in the world, with more than 750,000 deaths by 2020. With significant repercussions in public health, this pathology lacks biomarkers for early diagnosis, with endoscopy biopsy being the golden test for its detection. In the exploration of new strategies to control gastric cancer in recent years, liquid biopsy appears as a potential source of biomarkers using non-invasive procedures. MethodsHere we present the characterization of miRNAs contained in plasma-derived exosomes from patients with gastric cancer. Extracellular vesicles (EVs) were isolated using size-exclusion chromatography (SEC) and their characterization was performed by electron microscopy, protein expression, and nanoparticle analysis techniques. Total RNA from isolated exosomes was obtained for small RNA-seq analysis. ResultsTranscriptomic miRNA data were used to identify differentially expressed miRNAs between patients with benign and malignant gastric diseases, which resulted in a molecular signature of nine miRNAs, that were used in a regression model to classify individuals as either having benign or malignant disease. Further, we compared benign-malignant patients at different stages of gastric cancer, and we detected 15 differentially expressed miRNAs. Among these 15 miRNAs, miR-92a-3p, miR451a, and miR126-3p were identified as winners due to their clinical and regulatory relevance. ResultsOur results offer relevant information of a Colombian case study allowing us to propose three transcriptomic gastric cancer biomarkers in liquid biopsy. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=131 SRC="FIGDIR/small/538562v1_ufig1.gif" ALT="Figure 1"> View larger version (30K): org.highwire.dtl.DTLVardef@fca084org.highwire.dtl.DTLVardef@18e41cborg.highwire.dtl.DTLVardef@81ad97org.highwire.dtl.DTLVardef@1abb9b1_HPS_FORMAT_FIGEXP M_FIG C_FIG SummaryEVs are structures surrounded by a lipid bilayer that facilitate intercellular communication by transporting biomolecules commonly referred to as part of liquid biopsy. In this study, we examine the miRNAs contents of plasma isolated EVs from patients with both benign gastric diseases and gastric cancer to identify potential biomarkers for gastric cancer.

cancer biology↗

bioRxiv2023

Cryo-EM Structure of a Human LECT2 Amyloid Fibril Reveals a Network of Polar Ladders at its Core

ALECT2 is a type of systemic amyloidosis caused by deposition of the leukocyte cell-derived chemotaxin-2 (LECT2) protein in the form of fibrils. In ALECT2, LECT2 fibril deposits can be found in the glomerulus, resulting in renal failure. Affected patients lack effective treatment options outside of renal transplant or dialysis. While the structure of LECT2 in its globular form has been determined by X-ray crystallography, structures of LECT2 amyloid fibrils remain unknown. Using single particle cryo-EM, we now find that human LECT2 forms robust twisting fibrils with canonical amyloid features. At their core, LECT2 fibrils contain two mating protofilaments, the ordered core of each protofilament spans residues 55-75 of the LECT2 sequence. The overall geometry of the LECT2 fibril displays features in line with other pathogenic amyloids. Its core is tightly packed and stabilized by a network of hydrophobic contacts and hydrogen-bonded uncharged polar residues, while its outer surface displays several charged residues. The robustness of LECT2 fibril cores is illustrated by their limited dissolution in 3M urea and their persistence after treatment with proteinase K. As such, the LECT2 fibril structure presents a potential new target for treatments against ALECT2.

biochemistry↗

bioRxiv2022

Structure of a reversible amyloid fibril formed by the CPEB3 prion-like domain reveals a core sequence involved in translational regulation

The cytoplasmic polyadenylation element-binding protein 3 (CPEB3) is a prion-like RNA-binding polypeptide. As a functional prion, CPEB3 is thought to modulate protein synthesis at synapses and enable consolidation of long-term memory in neurons. Here, we report that the prion-like domain 1 of CPEB3 self-assembles into labile amyloid fibrils in vitro. A cryoEM structure of these fibrils reveals an ordered 48-residue core, spanning L103 to F151. CPEB3 constructs lacking this amyloidogenic segment form abnormal puncta in cells when compared to wild type CPEB3, with reduced localization in dormant p-bodies and increased localization in stress granules. Removal of the amyloid core segment in CPEB3 also abolishes its ability to regulate protein synthesis in neurons. Collectively, this evidence suggests that the newly identified amyloidogenic segment within the CPEB3 prion domain is important for its regulated aggregation in cells and suggest its involvement in regulating translational activity and potentially long-term memory formation.

biochemistry↗

bioRxiv2022

Sex-determining 3D regulatory hubs revealed by genome spatial auto-correlation analysis

Mammalian sex is determined by opposing networks of ovarian and testicular genes that are well characterized. However, its epigenetic regulation is still largely unknown, thus limiting our understanding of a fundamental process for species propagation. Here we explore the 3D chromatin landscape of sex determination in vivo, by profiling FACS-sorted embryonic mouse gonadal populations, prior and after sex determination, in both sexes. We integrate Hi-C with ChIP-seq experiments using METALoci, a novel genome spatial auto-correlation analysis that identifies 3D enhancer hubs across the genome. We uncover a prominent rewiring of chromatin interactions during sex determination, affecting the enhancer hubs of hundreds of genes that display temporal- and sex-specific expression. Moreover, the identification of the 3D enhancer hubs allows the reconstruction of regulatory networks, revealing key transcription factors involved in sex determination. By combining predictive approaches and validations in transgenic mice we identify a novel Fgf9 regulatory hub, deletion of which results in male-to-female sex reversal with the upregulation of ovarian-specific markers and the initiation of meiosis. Thus, spatial auto-correlation analysis is an effective strategy to identify regulatory networks associated to biological processes and to further characterize the functional role of the 3D genome.

developmental biology↗

bioRxiv2022

Lis1 relieves cytoplasmic dynein-1 auto-inhibition by acting as a molecular wedge

Cytoplasmic dynein-1 transports many intracellular cargos towards microtubule minus ends. Dynein is autoinhibited and undergoes conformational changes to form an active complex, consisting of one or two dynein dimers, the dynactin complex and activating adaptor(s)1,2. The Lissencephaly 1 gene, LIS1, is genetically linked to the dynein pathway from fungi to mammals and is mutated in patients with the neurodevelopmental disease lissencephaly3-5. Lis1 is required for active dynein complexes to form6-10, but how it does so is unclear. Here, we present a structure of two dynein motor domains with two Lis1 dimers wedged in-between. The contact sites between dynein and Lis1 in this structure, termed "Chi", are required for Lis1s regulation of dynein in Saccharomyces cerevisiae in vivo and the formation of active human dynein-dynactin- activating adaptor complexes in vitro. We propose that this structure represents an intermediate in dyneins activation pathway, revealing how Lis1 relieves dyneins autoinhibited state.

cell biology↗

bioRxiv2022

Autocrine Sfrp1 inhibits lung fibroblast invasion during transition to injury induced myofibroblasts

Fibroblast to myofibroblast conversion is a major driver of tissue remodeling in organ fibrosis. Several distinct lineages of fibroblasts support homeostatic tissue niche functions, yet, specific activation states and phenotypic trajectories of fibroblasts during injury and repair have remained unclear. Here, we combined spatial transcriptomics, longitudinal single-cell RNA-seq and genetic lineage tracing to study fibroblast fates during mouse lung regeneration. We discovered a transitional fibroblast state characterized by high Sfrp1 expression, derived from both Tcf21-Cre lineage positive and negative cells. Sfrp1+ cells appeared early after injury in peribronchiolar, adventitial and alveolar locations and preceded the emergence of myofibroblasts. We identified lineage specific paracrine signals and inferred converging transcriptional trajectories towards Sfrp1+ transitional fibroblasts and Cthrc1+ myofibroblasts. Tgf{beta}1 downregulated Sfrp1 in non-invasive transitional cells and induced their switch to an invasive Cthrc1+ myofibroblast identity. Finally, using loss of function studies we showed that autocrine Sfrp1 directly inhibits fibroblast invasion by regulating the RhoA pathway. In summary, our study reveals the convergence of spatially and transcriptionally distinct fibroblast lineages into transcriptionally uniform myofibroblasts and identifies Sfrp1 as an autocrine inhibitor of fibroblast invasion during early stages of fibrogenesis.

cell biology↗

bioRxiv2021

Fragment-based ab initio phasing of peptidic nanocrystals by MicroED

Microcrystal electron diffraction (MicroED) is transforming the visualization of molecules from nanocrystals, rendering their three-dimensional atomic structures from previously unamenable samples. Peptidic structures determined by MicroED include naturally occurring peptides, synthetic protein fragments and peptide-based natural products. However, as a diffraction method, MicroED is beholden to the phase problem, and its de novo determination of structures remains a challenge. ARCIMBOLDO, an automated, fragment-based approach to structure determination. It eliminates the need for atomic resolution, instead enforcing stereochemical constraints through libraries of small model fragments, and discerning congruent motifs in solution space to ensure validation. This approach expands the reach of MicroED to presently inaccessible peptidic structures including segments of human amyloids, and yeast and mammalian prions, and portends a more general phasing solution while limiting model bias for a wider set of chemical structures.

biochemistry↗

bioRxiv2020

DeepCOMBI: Explainable artificial intelligence for the analysis and discovery in genome-wide association studies

Deep learning algorithms have revolutionized data science in many fields by greatly improving prediction performances in comparison to conventional approaches. Recently, explainable artificial intelligence (XAI) has emerged as a novel area of research that goes beyond pure prediction improvement. Knowledge embodied in deep learning methodologies is extracted by interpreting their results. We investigate such explanations to explore the genetic architectures of phenotypes in genome-wide association studies. Instead of testing each position in the genome individually, the novel three-step algorithm, called DeepCOMBI, first trains a neural network for the classification of subjects into their respective phenotypes. Second, it explains the classifiers decisions by applying layerwise relevance propagation as one example from the pool of XAI techniques. The resulting importance scores are eventually used to determine a subset of most relevant locations for multiple hypothesis testing in the third step. The performance of DeepCOMBI in terms of power and precision is investigated on generated datasets and a 2007 WTCCC study. Verification of the latter is achieved by validating all findings with independent studies published up until 2020. DeepCOMBI is shown to outperform ordinary raw p-value thresholding as well as other baseline methods. Moreover, two novel disease associations (rs10889923 for hypertension and rs4769283 for type 1 diabetes) were identified.

bioinformatics↗