Biology subjects

Perera, J.

Publications and source records attributed to Perera, J..

14 recordsLinked to original sources

bioRxiv2026

Whole-genome 3D architectural screen reveals modulators of brain DNA structure

Three-dimensional (3D) genome architecture is the foundation of gene regulation, and plays a critical role in normal physiology and disease. However, our understanding of its biochemical determinants has long been limited by technology: imaging-based screens only profile a small number of loci, while sequencing-based studies rarely exceed 100 samples or conditions. Here we present "in-plate chromosome conformation capture" (Plate-C), a high-throughput, cost-effective platform that profiles thousands of whole-genome architectures in a day. Plate-C enabled the first chemical screen for whole-genome structural changes--profiling 2,956 samples from 834 conditions across 5 neuronal and glial types, accompanied by 6,081 single cells using "easy diploid chromosome conformation capture" (Easy Dip-C) and 200,893 single-cell transcriptomes. We discovered that diverse, dose/time-dependent, and cell type/species-specific modes of DNA structural changes can be rapidly induced by manipulating epigenetic (HDAC, BET), metabolic (mTOR), proteostatic (UPR), developmental (GSK3/Wnt, Hedgehog), immune (cGAS/STING), and neurotransmission pathways. To validate our finding in vivo, we demonstrated in newborn mice that HDAC inhibition drives brain-wide genome rewiring within hours, highly correlated with changes in vitro and inducing a latent structural and transcriptional state orthogonal to normal differentiation. By enabling massively parallel profiling of whole-genome structures, Plate-C paves the way for systematic discovery of DNA folding principles to better understand and engineer the human genome in 3D.

genomics↗

bioRxiv2026

AI-Guided CRISPR Screen Accelerates Discovery of New Drug Targets

Psoriasis affects over 125 million people worldwide, yet the mechanistic understanding of keratinocyte-driven inflammation remains incomplete, limiting therapeutic innovation beyond costly systemic biologics that are prone to side effects. Here, we performed the first genome-wide CRISPR knockout screen in primary human adult epidermal keratinocytes to systematically identify regulators of IL-17 receptor A (IL17RA), a central node in psoriatic inflammation. To prioritize therapeutically tractable targets from over 19,000 screened genes, we integrated a large language model - VirtualCRISPR - trained on functional genomics data, identifying arachidonate 5-lipoxygenase (ALOX5) and oxytocin receptor (OXTR) as high-confidence novel hits with minimal prior association with psoriasis. Multi-omics validation revealed that ALOX5 and OXTR regulate IL17RA expression through distinct signaling pathways - ALOX5 through lipid mediators that stabilize the receptor at the cell surface, and OXTR through calcium signaling that reprograms cellular metabolism. Topical delivery of their inhibitors Zileuton (ALOX5) and Cligosiban (OXTR) exhibited therapeutic efficacy comparable to systemic anti-IL17RA antibody in the imiquimod-induced psoriasis model, suppressing pathogenic Th17/Tc17 responses, polarizing macrophages toward anti-inflammatory phenotypes, and normalizing epidermal hyperproliferation. Proteomic profiling in human 3D organotypic skin and murine models confirmed on-target pharmacology and revealed convergent suppression of neutrophil-keratinocyte inflammatory circuits. The use of VirtualCRISPR significantly shortened the timescale from screen to the identification of druggable hits with robust validation, and this work establishes a blueprint for integrating AI-driven target prioritization with functional genomics to accelerate therapeutic discovery.

genomics↗

bioRxiv2026

Virtual Cells Need Context, Not Just Scale

The intersection of AI and biology has entered a phase of explosive growth, driven by the ambition to build "Virtual Cells" or computational models capable of predicting cellular responses to any perturbation. Following the success of structural biology (e.g., AlphaFold) and large language models, the field has converged on training massive, high-capacity models on large-scale single-cell data. This position paper argues that scaling model capacity is insufficient to solve the Virtual Cell problem because the primary failure mode is a lack of adequate coverage over diverse biological contexts, not insufficient model expressivity. We support this claim by reviewing recent studies showing that simple baselines perform on par with sophisticated architectures within a given biological context, and current models fail to consistently generalize across contexts. We connect this finding to the causal inference literature on transportability and contrast it with domains where scaling has succeeded. We substantiate our argument through analysis of a state-of-the-art model on a 22-million-cell immunology dataset. We conclude that the community faces a causal transport problem that cannot be solved by accumulating more data from the same distributions. Instead, we argue that contextual diversity and causal representation learning deserve increased emphasis, complementing ongoing scaling of model capacity and data volume.

genomics↗

bioRxiv2026

DecoderTCR: Compositional Pretraining and Entropy-Guided Decoding for TCR-pMHC Interactions

Modeling recognition between T-cell receptors (TCRs) and peptide-MHC (pMHC) complexes is a fundamental challenge in computational immunology, constrained by sparse paired interaction data relative to abundant unpaired sequences. We introduce DecoderTCR, a masked language model framework that addresses this through two contributions: (1) a compositional continual pre-training curriculum that learns component representations from marginal data before refining cross-chain dependencies from limited pairs, and (2) Iterative Entropy-Guided Refinement (IEGR), a non-autoregressive decoding algorithm that resolves high-confidence positions first to provide context for uncertain regions. On held-out benchmarks, DecoderTCR achieves 0.96 AUROC for zero-shot pMHC binding prediction and 0.76 AUROC for epitope-specific TCR recognition, approaching supervised baselines without epitope-specific training. Learned representations recover structural contacts without coordinate supervision, and generated sequences exhibit realistic recombination statistics. Experimental validation reveals a prediction-generation gap: strong discrimination does not yet yield reliable generation, highlighting an open challenge for the field.

bioinformatics↗

bioRxiv2025

Lymphotoxin-driven cancer cell eradication by tumoricidal CD8+ tumor-infiltrating lymphocytes

Tumor-infiltrating lymphocyte (TIL) therapy is FDA-approved for patients with treatment-resistant advanced melanoma, but the TIL subpopulations critical for tumor eradication remains incompletely understood. Using patient-derived TIL-melanoma co-cultures, we identified and characterized a novel subset of CD8+ TIL, capable of class I HLA-independent cancer cell lysis. The lymphotoxin {beta} receptor (LT{beta}R) and interferon (IFN) sensing pathways were nominated as key determinants of TIL-mediated cancer cell killing from a whole-genome, loss-of-function CRISPR screen. Validation studies confirmed that dual LT{beta}R and IFN sensing is necessary and sufficient for cancer cell lysis, and that expanded CD8+ TIL express high lymphotoxin {beta} (LTB) and upregulate lymphotoxin (LTA) upon coculture with cancer cells. Leveraging paired scRNA-seq and scTCR-seq data, we confirmed that enrichment of LTB+CD8+ T cells is associated with clinical response to TIL, and that LTB+CD8+ TIL are expanded from putative neoantigen-reactive, LTBlo CD8+ T cells in resected tumors. SignificanceWe have uncovered a previously unrecognized mechanism of TIL-mediated tumor eradication, providing mechanistic insights into the role of LTBR/IFN signaling in TIL-mediated cancer cell killing, and potentially offering insights into novel strategies to isolate, enrich, and expand tumoricidal TIL or augment specific TIL functions to enhance tumor control.

cancer biology↗

bioRxiv2025

Enrichment of a CD4-CD8- NK-like cytotoxic Vδ1/3 T cell subset in tuberculosis disease

Tuberculosis (TB), caused by Mycobacterium tuberculosis, remains a leading cause of global morbidity and mortality. Although gamma-delta ({gamma}{delta}) T cells have increasingly been suggested to contribute to the TB immune response, quantitative and qualitative differences in this immune cell compartment between healthy and TB diseased individuals are not well-characterized. In this study, we used single-cell RNA sequencing to provide a high-resolution characterization of CD4-CD8- {gamma}{delta} T cells in peripheral blood across healthy Mtb-non-sensitized, healthy Mtb-sensitized, and TB disease pre-/post-treatment cohorts. We found upregulation of an activated and cytotoxic gene signature in {gamma}{delta} T cells of TB disease compared to both healthy cohorts. Strikingly, these differences persisted through one year following diagnosis of TB disease (corresponding to six months after completion of anti-TB therapy). We found that these transcriptomic differences were largely mediated by an NK-like cytotoxic V{delta}1 and V{delta}3 subset that was enriched in TB disease, with a unique V{delta}3 TCR gene usage. Our findings suggest long-lasting changes in the CD4-CD8- {gamma}{delta} T cell compartment and highlight V{delta}3 cells, a previously underappreciated {gamma}{delta} T cell subset, as potentially important in TB.

immunology↗

bioRxiv2025

Structural variants are enriched in deleterious visible phenotypes in Drosophila

Genome structural variants (SVs) comprise a sizable portion of functionally important genetic variation in all organisms; yet, many SVs evade discovery using short reads. While long-read sequencing can find the hidden SVs, the role of SVs in variation in organismal traits remains largely unclear. To address this gap, we investigate the molecular basis of 50 classical phenotypes in 11 Drosophila melanogaster strains using highly contiguous de novo genome assemblies generated with Oxford Nanopore long reads. These assemblies enabled the creation of a pangenome graph containing comprehensive, nucleotide-resolution maps of SVs, including complex rearrangements such as the interchromosomal inverted duplication Dp(2;4)eyD and large tandem duplications at the Bar locus. We uncovered new candidate causal mutations for 15 phenotypes and new molecular alleles for 2 mutations comprising tandem duplications, transposable element (TE) insertions, and indels. For example, we mapped the tarsal joint defect AblpeyD to an 8 kb Roo retrotransposon insertion into an intergenic enhancer, a finding validated via CRISPR-Cas9. The wing vein phenotype plexus (px1) was linked to a 1.5 kb partial tandem gene duplication, and the century-old Curved (c1) wing phenotype was linked to a 7.5 kb DM412 retrotransposon inserted into the coding sequence of the muscle protein gene Strn-Mlck. We also unveiled 8 SV alleles of previously identified causal genes, including previously uncharacterized SVs underlying the extensively studied white and yellow phenotypes. Overall, 67.4% of the genes causing phenotypic changes harbored candidate SVs over 100 bp, whereas only 28% is expected based on euchromatic SVs. Our data, based on the 50 Drosophila phenotypes, 44 of which are strongly deleterious, suggests a disproportionately larger contribution of SVs to deleterious changes in visible phenotypes in Drosophila.

genomics↗

bioRxiv2025

Peptide Driven Identification of TCRs (PDI-TCR) reveals dynamics and phenotypes of CD4 T cells in tuberculosis

Assigning antigen specificity to T cell receptor (TCR) sequences is challenging due to the TCR repertoires diversity and the complexity of TCR:antigen recognition. We developed the Peptide-Driven Identification of TCRs (PDI-TCR) assay that combines in vitro expansion of cells with peptide pools, bulk TCR sequencing, and statistical analysis to identify antigen-specific TCRs from human blood. A key feature of PDI-TCR is the ability to distinguish true antigen-specific TCR clonotypes from TCRs associated with unspecific bystander activation by comparing responses to non-overlapping peptide pools. We applied PDI-TCR to Tuberculosis (TB) patients, sampling blood at diagnosis and throughout treatment, and Mycobacterium tuberculosis (Mtb)-sensitized healthy individuals (IGRA+). We identified hundreds of Mtb-specific TCRs, as well as unspecific TCRs, and characterized their phenotype in each cohort by single-cell RNA sequencing ex vivo. Mtb-specific T cells were highly diverse, with short-lived effector phenotypes only present in TB at diagnosis, while memory phenotypes were maintained through treatment. In contrast, unspecific expanded T cells were more clonally restricted, had a cytotoxic phenotype, and were maintained throughout treatment. This showcases PDI-TCR as a powerful tool for identifying antigen-specific TCRs, which enables direct ex vivo identification and monitoring of antigen-specific T cells.

immunology↗

bioRxiv2025

Sialylated CD43 is a glyco-immune checkpoint for macrophage phagocytosis

Macrophages in the tumor microenvironment exert potent anti-tumorigenic activity through phagocytosis. Yet therapeutics that enhance macrophage phagocytosis have not improved outcomes in clinical trials for patients with acute myeloid leukemia (AML) or myelodysplastic syndrome (MDS). To systematically identify regulators of phagocytosis, we performed genome-scale CRISPR knockout screens in human leukemia cells co-cultured with human monocyte-derived macrophages. Surprisingly, we found that whereas the classic "dont eat me" signal CD47 inhibited mouse macrophages, it did not inhibit phagocytosis by human macrophages. In contrast, the O-linked glycosylation and sialylation pathways were strong negative regulators of phagocytosis. In AML, the cell surface O-linked glycoprotein CD43 was the major effector of the O-linked glycosylation and sialylation pathways. Genetic deletion or antibody blockade of CD43 enhanced macrophage phagocytosis. This work highlights the importance of using human platforms to identify immune checkpoints, and nominates CD43 as a glyco-immune regulator of human macrophage phagocytosis.

cancer biology↗

bioRxiv2024

Pep2Vec: An Interpretable Model for Peptide-MHC Presentation Prediction and Contaminant Identification in Ligandome Datasets

As personalized cancer vaccines advance, precise modeling of antigen presentation by MHC class I and II is crucial. High-quality training data is essential for clinical models. Existing deep learning models focus on prediction performance but lack interpretability. We introduce Pep2Vec, a modular, transformer-based model trained on MHC I and II ligandome data, transforming input sequences into interpretable vectors. This approach integrates source protein features and elucidates the source of its performance gains, revealing regions that correlate with gene expression and protein-protein interactions. Pep2Vecs peptide latent space shows relationships between peptides of varying MHC class, allotype, lengths, and submotifs. This enables identifying four major contaminant types, constituting 5.0% of our data. Pep2Vec enhances MHC presentation prediction, achieving higher average precision on our presentation test set and immunogenicity datasets than existing models, and reducing contaminant-like peptide recommendations. Pep2Vec addresses a critical need for the development of more precise and effective applications of peptide MHC models, such as for cancer vaccines and antibody deimmunization.

immunology↗

bioRxiv2023

Graph-pMHC: Graph Neural Network Approach to MHC Class II Peptide Presentation and Antibody Immunogenicity

Antigen presentation on MHC Class II (pMHCII presentation) plays an essential role in the adaptive immune response to extracellular pathogens and cancerous cells. But it can also reduce the efficacy of large-molecule drugs by triggering an anti-drug response. Significant progress has been made in pMHCII presentation modeling due to the collection of large-scale pMHC mass spectrometry datasets (ligandomes) and advances in deep machine learning. Here, we develop graph-pMHC, a graph neural network approach to predict pMHCII presentation. We derive adjacency matrices for pMHCII using Alphafold2-multimer, and address the peptide-MHC binding groove alignment problem with a simple graph enumeration strategy. We demonstrate that graph-pMHC dramatically outperforms methods with suboptimal inductive biases, such as the multilayer-perceptron-based NetMHCIIan-4.0 (+22.84% average precision). Finally, we create an antibody drug immunogenicity dataset from clinical trial data, and develop a method for measuring anti-antibody immunogenicity risk using pMHCII presentation models. In comparison with BioPhis Sapiens score, a deep learning based measure of the humanness of an antibody drug, our strategy achieves a 7.14% ROC AUC improvement in predicting antibody drug immunogenicity.

immunology↗

bioRxiv2022

Effects of Aging on Cortical Representations of Continuous Speech

Understanding speech in a noisy environment is crucial in day-to-day interactions, and yet becomes more challenging with age, even for healthy aging. Age-related changes in the neural mechanisms that enable speech-in-noise listening have been investigated previously; however, the extent to which age affects the timing and fidelity of encoding of target and interfering speech streams are not well understood. Using magnetoencephalography (MEG), we investigated how continuous speech is represented in auditory cortex in the presence of interfering speech, in younger and older adults. Cortical representations were obtained from neural responses that time-locked to the speech envelopes using speech envelope reconstruction and temporal response functions (TRFs). TRFs showed three prominent peaks corresponding to auditory cortical processing stages: early ([~]50 ms), middle ([~]100 ms) and late ([~]200 ms). Older adults showed exaggerated speech envelope representations compared to younger adults. Temporal analysis revealed both that the age-related exaggeration starts as early as [~]50 ms, and that older adults needed a substantially longer integration time window to achieve their better reconstruction of the speech envelope. As expected, with increased speech masking, envelope reconstruction for the attended talker decreased and all three TRF peaks were delayed, with aging contributing additionally to the reduction. Interestingly, for older adults the late peak was delayed, suggesting that this late peak may receive contributions from multiple sources. Together these results suggest that there are several mechanisms at play compensating for age-related temporal processing deficits at several stages, but which are not able to fully reestablish unimpaired speech perception. NEW & NOTEWORTHYWe observed age-related changes in cortical temporal processing of continuous speech that may be related to older adults difficulty understanding speech in noise. These changes occur in both timing and strength of the speech representations at different cortical processing stages, and depend on both noise condition and selective attention. Critically, their dependency on noise condition changes dramatically among the early, middle, and late cortical processing stages, underscoring how aging differentially affects these stages.

neuroscience↗

bioRxiv2021

Palaeogenomic analysis of black rat (Rattus rattus) reveals multiple European introductions associated with human economic history

The distribution of the black rat (Rattus rattus) has been heavily influenced by its association with humans. The dispersal history of this non-native commensal rodent across Europe, however, remains poorly understood, and different introductions may have occurred during the Roman and medieval periods. Here, in order to reconstruct the population history of European black rats, we generated a de novo genome assembly of the black rat, 67 ancient black rat mitogenomes and 36 ancient nuclear genomes from sites spanning the 1st-17th centuries CE in Europe and North Africa. Analyses of mitochondrial DNA confirm that black rats were introduced into the Mediterranean and Europe from Southwest Asia. Genomic analyses of the ancient rats reveal a population turnover in temperate Europe between the 6th and 10th centuries CE, coincident with an archaeologically attested decline in the black rat population. The near disappearance and re-emergence of black rats in Europe may have been the result of the breakdown of the Roman Empire, the First Plague Pandemic, and/or post-Roman climatic cooling.

evolutionary biology↗

bioRxiv2019

Isoniazid resistance in Mycobacterium tuberculosis is a heterogeneous phenotype comprised of overlapping MIC distributions with different underlying resistance mechanisms

MIC testing using the BACTEC 960 MGIT system of 70 phylogenetically diverse, isoniazid-resistant clinical strains of Mycobacterium tuberculosis revealed a complex pattern of overlapping MIC distributions. Whole-genome sequencing could explain most of the level of resistance observed. The MIC distribution of strains with only inhA promoter mutations was split by the current concentration that is endorsed by the Clinical Laboratory Standards Institute to detect low-level resistance to isoniazid and is therefore likely not optimally set.

microbiology↗