Biology subjects

Warrell, J.

Publications and source records attributed to Warrell, J..

11 recordsLinked to original sources

bioRxiv2026

Generative AI Guided Design of High-Affinity T cell Receptors

Developing T cell receptors (TCRs) with sufficiently high affinity for tumor antigens (TAs) remains a fundamental challenge in TCR-T immunotherapy. Experimental methods such as affinity maturation and high-throughput screening have enabled the identification of TCRs with enhanced activity. However, their efficiency is often constrained by limited throughput, insufficient coverage, and the generally lower affinities of naturally occurring TCRs toward TAs. To address these challenges, we present TCRPPO2, an integrated AI-driven, in silico affinity maturation framework for peptide-specific TCR optimization. Using reinforcement learning, TCRPPO2 learns mutation policies that iteratively enhance the TCR binding affinity to the target peptide, derived from predictive models trained on carefully curated interaction data. The model is further augmented by a generative AI critic model that discourages implausible designs to ensure the biophysical validity. The designs are further screened by robust post-screening methods that leverage diverse functional annotations and physical prior knowledge. We applied TCRPPO2 to the clinically relevant MART-1 antigen and experimentally validated the designed candidates in Jurkat cell-based functional assays. Among the five engineered TCRs, all of which demonstrated positive cellular responses, three showed significantly increased activities relative to their templates and one showed substantial enhancement. These functional gains were consistent with more favorable interaction energy from structural and physical modeling. Together, our results support a generalizable paradigm for TCR engineering, in which learned mutation policies can efficiently navigate the peptide-specific binding landscape of TCRs and propose biologically enhanced candidates without explicit structural supervision, offering a practical route for early-stage computational TCR optimization for challenging tumor antigens.

immunology↗

bioRxiv2024

Predicting Spatially Resolved Gene Expression via Tissue Morphology using Adaptive Spatial GNNs

MotivationSpatial transcriptomics technologies, which generate a spatial map of gene activity, can deepen the understanding of tissue architecture and its molecular underpinnings in health and disease. However, the high cost makes these technologies difficult to use in practice. Histological images co-registered with targeted tissues are more affordable and routinely generated in many research and clinical studies. Hence, predicting spatial gene expression from the morphological clues embedded in tissue histological images, provides a scalable alternative approach to decoding tissue complexity. ResultsHere, we present a graph neural network based framework to predict the spatial expression of highly expressed genes from tissue histological images. Extensive experiments on two separate breast cancer data cohorts demonstrate that our method improves the prediction performance compared to the state-of-the-art, and that our model can be used to better delineate spatial domains of biological interest. Availabilityhttps://github.com/song0309/asGNN/

bioinformatics↗

bioRxiv2024

Single-cell genomics and regulatory networks for 388 human brains

Single-cell genomics is a powerful tool for studying heterogeneous tissues such as the brain. Yet, little is understood about how genetic variants influence cell-level gene expression. Addressing this, we uniformly processed single-nuclei, multi-omics datasets into a resource comprising >2.8M nuclei from the prefrontal cortex across 388 individuals. For 28 cell types, we assessed population-level variation in expression and chromatin across gene families and drug targets. We identified >550K cell-type-specific regulatory elements and >1.4M single-cell expression-quantitative-trait loci, which we used to build cell-type regulatory and cell-to-cell communication networks. These networks manifest cellular changes in aging and neuropsychiatric disorders. We further constructed an integrative model accurately imputing single-cell expression and simulating perturbations; the model prioritized [~]250 disease-risk genes and drug targets with associated cell types. Summary Figure O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=95 SRC="FIGDIR/small/585576v2_ufig1.gif" ALT="Figure 1"> View larger version (33K): org.highwire.dtl.DTLVardef@70b24org.highwire.dtl.DTLVardef@6c9eddorg.highwire.dtl.DTLVardef@131fae2org.highwire.dtl.DTLVardef@4e0f1e_HPS_FORMAT_FIGEXP M_FIG C_FIG

genomics↗

bioRxiv2024

A Variational Graph Partitioning Approach to Modeling Protein Liquid-liquid Phase Separation

Graph Neural Network (GNN)s have emerged as a powerful general-purpose tool for representation learning across many domains. Their efficacy often depends on having an optimal underlying graph for prediction. In many cases, the most relevant information comes from specific subgraphs. In this work, we introduce a novel GNN architecture, called Graph Partitioned GNN (GP-GNN), designed to partition graphs during the prediction process, thereby focusing attention on subgraphs containing the most relevant information. Our approach jointly learns task-dependent graph partitions and representations, making it particularly effective for predictive tasks where critical features may reside within initially unidentified subgraphs. Protein Liquid-Liquid Phase Separation (LLPS) is an important physical problem well-suited to our architecture, primarily because protein sub-domains called intrinsically disordered regions (IDR)s are known to play a crucial role in the phase separation process. LLPS plays an essential role in cellular processes and is known to be associated with various diseases (e.g ., neurodegenerative diseases and cancer). However, our ability to accurately predict which proteins undergo LLPS remains limited. In this study, we demonstrate how GP-GNN can be utilized to accurately predict LLPS by partitioning protein graphs into task-relevant subgraphs, such as those highlighting IDRs. Our model not only achieves state-of-the-art accuracy in predicting LLPS for both regulator and scaffold proteins but also offers valuable biological insights, that can be used to guide further downstream investigations. Notably, upon examining subgraphs identified by the GP-GNN, we show these are consistent with annotated IDRs.

bioinformatics↗

bioRxiv2022

Insights from Incorporating Quantum Computing into Drug Design Workflows

While many quantum computing (QC) methods promise theoretical advantages over classical counterparts, quantum hardware remains limited. Exploiting near-term QC in computer-aided drug design (CADD) thus requires judicious partitioning between classical and quantum calculations. We present HypaCADD, a hybrid classical-quantum workflow for finding ligands binding to proteins, while accounting for genetic mutations. We explicitly identify modules of our drug design workflow currently amenable to replacement by QC: non-intuitively, we identify the mutation-impact predictor as the best candidate. HypaCADD thus combines classical docking and molecular dynamics with quantum machine learning (QML) to infer the impact of mutations. We present a case study with the SARS-CoV-2 protease and associated mutants. We map a classical machine-learning module onto QC, using a neural network constructed from qubit-rotation gates. We have implemented this in simulation and on two commercial quantum computers. We find that the QML models can perform on par with, if not better than, classical baselines. In summary, HypaCADD offers a successful strategy for leveraging QC for CADD.

bioinformatics↗

bioRxiv2020

Compression-based Network Interpretability Schemes

Deep learning methods have achieved state-of-the-art performance in many domains of artificial intelligence, but are typically hard to interpret. Network interpretation is important for multiple reasons, including knowledge discovery, hypothesis generation, fairness and establishing trust. Model transformations provide a general approach to interpreting a trained network post-hoc: the network is approximated by a model, which is typically compressed, whose structure can be more easily interpreted in some way (we call such approaches interpretability schemes). However, the relationship between compression and interpretation has not been fully explored: How much should a network be compressed for optimal extraction of interpretable information? Should compression be combined with other criteria when selecting model transformations? We investigate these issues using two different compression-based schemes, which aim to extract orthogonal kinds of information, pertaining to feature and data instance-based groupings respectively. The first (rank projection trees) uses a structured sparsification method such that nested groups of features can be extracted having potential joint interactions. The second (cascaded network decomposition) splits a network into a cascade of simpler networks, allowing groups of training instances with similar characteristics to be extracted at each stage of the cascade. We use predictive tasks in cancer and psychiatric genomics to assess the ability of these approaches to extract informative feature and data-point groupings from trained networks. We show that the generalization error of a network provides an indicator of the quality of the information extracted; further we derive PAC-Bayes generalization bounds for both schemes, which we show can be used as proxy indicators, and can thus provide a criterion for selecting the optimal compression. Finally, we show that the PAC-Bayes framework can be naturally modified to incorporate additional criteria alongside compression, such as prior knowledge based on previous models, which can enhance interpretable model selection.

bioinformatics↗

bioRxiv2020

Latent Evolutionary Signatures: A General Framework for Analyzing Music and Cultural Evolution

Cultural processes of change bear many resemblances to biological evolution. The underlying units of non-biological evolution have, however, remained elusive, especially in the domain of music. Here, we introduce a general framework to jointly identify underlying units and their associated evolutionary processes. We model musical styles and principles of organization in dimensions such as harmony and form as following an evolutionary process. Furthermore, we propose that such processes can be identified by extracting latent evolutionary signatures from musical corpora, analogous to identifying mutational signatures in genomics. These signatures provide a latent embedding for each song or musical piece. We develop a deep generative architecture for our model, which can be viewed as a type of Variational Autoencoder with an evolutionary prior constraining the latent space; specifically, the embeddings for each song are tied together via an energy-based prior, which encourages songs close in evolutionary space to share similar representations. As illustration, we analyze songs from the McGill Billboard dataset. We find frequent chord transitions and formal repetition schemes and identify latent evolutionary signatures related to these features. Finally, we show that the latent evolutionary representations learned by our model outperform non-evolutionary representations in such tasks as period and genre prediction.

evolutionary biology↗

bioRxiv2019

Cyclic and Multilevel Causation in Evolutionary Processes

Many models of evolution are implicitly causal processes. Features such as causal feedback between evolutionary variables and evolutionary processes acting at multiple levels, though, mean that conventional causal models miss important phenomena. We develop here a general theoretical framework for analyzing evolutionary processes drawing on recent approaches to causal modeling developed in the machine-learning literature, which have extended Pearls do-calculus to incorporate cyclic causal interactions and multilevel causation. We also develop information-theoretic notions necessary to analyze causal information dynamics in our framework, introducing a causal generalization of the Partial Information Decomposition framework. We show how our causal framework helps to clarify conceptual issues in the contexts of complex trait analysis and cancer genetics, including assigning variation in an observed trait to genetic, epigenetic and environmental sources in the presence of epigenetic and environmental feedback processes, and variation in fitness to mutation processes in cancer using a multilevel causal model respectively, as well as relating causally-induced to observed variation in these variables via information theoretic bounds. In the process, we introduce a general class of multilevel causal evolutionary processes which connect evolutionary processes at multiple levels via coarse-graining relationships. Further, we show how a range of fitness models can be formulated in our framework, as well as a causal analog of Prices equation (generalizing the probabilistic Rice equation), clarifying the relationships between realized/probabilistic fitness and direct/indirect selection. Finally, we consider the potential relevance of our framework to foundational issues in biology and evolution, including supervenience, multilevel selection and individuality. Particularly, we argue that our class of multilevel causal evolutionary processes, in conjunction with a minimum description length principle, provides a conceptual framework in which identification of multiple levels of selection may be reduced to a model selection problem.

evolutionary biology↗

bioRxiv2019

Estimating growth patterns and driver effects in tumor evolution from individual samples

Evolving tumors accumulate thousands of mutations. Technological advances have enabled whole genome sequencing of these mutations in large cohorts, such as those from the Pancancer Analysis of Whole Genomes (PCAWG) Consortium. The resulting data explosion has led to many methods for detecting cancer drivers through mutational recurrence and deviation from background mutation rates. However, these methods require a large cohort and underperform when recurrence is low. An alternate approach involves harnessing the variant allele frequency (VAF) of mutations in the population of tumor cells in a single individual. Moreover, ultra-deep sequencing of tumors, which is now possible, allows for particularly accurate VAF measurements, and recent studies have begun to use these to determine evolutionary trajectories and quantify subclonal selection. Here, we developed a method that quantifies tumor growth and driver effects for individual samples based solely on the VAF spectrum. Drivers introduce a perturbation into this spectrum, and our method uses the frequency of \"hitchhiking\" mutations preceding a driver to measure this perturbation. Specifically, our method applies various growth models to identify periods of positive/negative growth, the genomic regions associated with them, and the presence and effect of putative drivers. To validate our method, we first used simulation models to successfully approximate the timing and size of a drivers effect. Then, we tested our method on 993 linear tumors (i.e. those with linear subclonal expansion, where each parent-subclone has one child) from the PCAWG Consortium and found that the identified periods of positive growth are associated with drivers previously highlighted via recurrence by the PCAWG consortium. Finally, we applied our method to an ultra-deep sequenced AML tumor and identified known cancer genes and additional driver candidates. In summary, our method presents opportunities for personalized diagnosis using deep sequenced whole genome data from an individual.

cancer biology↗

bioRxiv2019

An integrative ENCODE resource for cancer genomics

ENCODE comprises thousands of functional genomics datasets, and the encyclopedia covers hundreds of cell types, providing a universal annotation for genome interpretation. However, for particular applications, it may be advantageous to use a customized annotation. Here, we develop such a custom annotation by leveraging advanced assays, such as eCLIP, Hi-C, and whole-genome STARR-seq on a number of data-rich ENCODE cell types. A key aspect of this annotation is comprehensive and experimentally derived networks of both transcription factors and RNA-binding proteins (TFs and RBPs). Cancer, a disease of system-wide dysregulation, is an ideal application for such a network-based annotation. Specifically, for cancer-associated cell types, we put regulators into hierarchies and measure their network change (rewiring) during oncogenesis. We also extensively survey TF-RBP crosstalk, highlighting how SUB1, a previously uncharacterized RBP, drives aberrant tumor expression and amplifies the effect of MYC, a well-known oncogenic TF. Furthermore, we show how our annotation allows us to place oncogenic transformations in the context of a broad cell space; here, many normal-to-tumor transitions move towards a stem-like state, while oncogene knockdowns show an opposing trend. Finally, we organize the resource into a coherent workflow to prioritize key elements and variants, in addition to regulators. We showcase the application of this prioritization to somatic burdening, cancer differential expression and GWAS. Targeted validations of the prioritized regulators, elements and variants using siRNA knockdowns, CRISPR-based editing, and luciferase assays demonstrate the value of the ENCODE resource.

genomics↗

bioRxiv2018

Passenger mutations in 2500 cancer genomes: Overall molecular functional impact and consequences

The Pan-cancer Analysis of Whole Genomes (PCAWG) project provides an unprecedented opportunity to comprehensively characterize a vast set of uniformly annotated coding and non-coding mutations present in thousands of cancer genomes. Classical models of cancer progression posit that only a small number of these mutations strongly drive tumor progression and that the remaining ones (termed \"putative passengers\") are inconsequential for tumorigenesis. In this study, we leveraged the comprehensive variant data from PCAWG to ascertain the molecular functional impact of each variant. The impact distribution of PCAWG mutations shows that, in addition to high- and low-impact mutations, there is a group of medium-impact putative passengers predicted to influence gene activity. Moreover, the predicted impact relates to the underlying mutational signature: different signatures confer divergent impact, differentially affecting distinct regulatory subsystems and gene categories. We also find that impact varies based on subclonal architecture (i.e., early vs. late mutations) and can be related to patient survival. Finally, we note that insufficient power due to limited cohort sizes precludes identification of weak drivers using standard recurrence-based approaches. To address this, we adapted an additive effects model derived from complex trait studies to show that aggregating the impact of putative passenger variants (i.e. including yet undetected weak drivers) provides significant predictability for cancer phenotypes beyond the PCAWG identified driver mutations (12.5% additive variance). Furthermore, this framework allowed us to estimate the frequency of potential weak driver mutations in the subset of PCAWG samples lacking well-characterized driver alterations.

genomics↗