Biology subjects

Atwal, G. S.

Publications and source records attributed to Atwal, G. S..

14 recordsLinked to original sources

bioRxiv2024

PROPERMAB: an integrative framework for in silico prediction of antibody developability using machine learning

Selection of lead therapeutic molecules is often driven predominantly by pharmacological efficacy and safety. Candidate developability, such as biophysical properties that affect the formulation of the molecule into a product, is usually evaluated only toward the end of the drug development pipeline. The ability to evaluate developability properties early in the process of antibody therapeutic development could accelerate the timeline from discovery to clinic and save considerable resources. In silico predictive approaches, such as machine learning models, which map molecules to predictions of developability properties could offer a cost-effective and high-throughput alternative to experiments for antibody developability assessment. We developed a computational framework, PROPERMAB, for large-scale and efficient in silico prediction of developability properties for monoclonal antibodies, using custom molecular features and machine learning modeling. We demonstrate the power of PROPERMAB by using it to develop models to predict antibody hydrophobic interaction chromatography retention time and high-concentration viscosity. We further show that structure-derived features can be rapidly and accurately predicted directly from sequences by pre-training simple models for molecular features, thus providing the ability to scale these approaches to repertoire-scale sequence datasets.

bioinformatics↗

bioRxiv2024

Optimizing murine sample sizes for RNA-seq studies revealed from large-scale comparative analysis

Determining the appropriate sample size (N) for comparative biological experiments is critical for obtaining reliable results. In order to determine the N, the usual approach is to perform a power calculation, which involves understanding the variability between samples and the expected effect size. Here, we focused on bulk RNA-seq experiments, which have become ubiquitous in biology, but which have many unknown or difficult to estimate parameters, and so the required analyses to determine the minimum N is typically lacking. We therefore performed two N=30 profiling studies between wild-type mice and mice in which one copy of a gene had been deleted, to determine how many mice would be required to minimize false positives and to maximize true discoveries found in the N of 30 experiment. Results from experiments with N=4 or less are shown to be highly misleading, given the substantial false positive rate, and the lack of discovery of genes later found with higher N. For a cut-off of 2-fold expression differences, we found that an N of 6-7 mice was required to consistently decrease the false positive rate to below 50%, and that "more is always better" when it came to discovery rates - an N of 8-12 is significantly better in lowering the false positive rate. A common method to reduce false discovery rate in underpowered experiments is to raise the fold cutoff or increase the stringency of the P-value and include only highly perturbed, highly significant genes. We show that while this strategy is no substitute for increasing the N of the experiment, because it results in consistently inflated effect sizes and a substantial drop in sensitivity. These data should be helpful to others in choosing their Ns, since its often not practical to do such large studies for every mouse model.

bioinformatics↗

bioRxiv2023

Image guided construction of a common coordinate framework for spatial transcriptome data

Spatial transcriptomics is a powerful technology for high-resolution mapping of gene expression in tissue samples, enabling a molecular level understanding of tissue architecture. The acquisition entails dissecting and profiling micron-thick tissue slices, with multiple slices often needed for a comprehensive study. However, the lack of a common coordinate framework (CCF) among slices, due to slicing and displacement variations, can hinder data analysis, making data comparison and integration challenging, and potentially compromising analysis accuracy. Here we present a deep learning algorithm STaCker that unifies the coordinates of transcriptomic slices via an image registration process. STaCker derives a composite image representation by integrating tissue image and gene expressions that are transformed to be resilient to noise and batch effects. Trained exclusively on diverse synthetic data, STaCker overcomes the training data scarcity and is applicable to any tissue type. Its performance on various benchmarking datasets shows a significant increase in spatial concordance in aligned slices, surpassing existing methods. STaCker also successfully harmonizes multiple real spatial transcriptome datasets. These results indicate that STaCker is a valuable computational tool for constructing a CCF with spatial transcriptome data.

bioinformatics↗

bioRxiv2023

Designing meaningful continuous representations of T cell receptor sequences with deep generative models

T Cell Receptor (TCR) antigen binding underlies a key mechanism of the adaptive immune response yet the vast diversity of TCRs and the complexity of protein interactions limits our ability to build useful low dimensional representations of TCRs. To address the current limitations in TCR analysis we develop a capacity-controlled disentangling variational autoencoder trained using a dataset of approximately 100 million TCR sequences, that we name TCR-VALID. We design TCR-VALID such that the model representations are low-dimensional, continuous, disentangled, and sufficiently informative to provide high-quality TCR sequence de novo generation. We thoroughly quantify these properties of the representations, providing a framework for future protein representation learning in low dimensions. The continuity of TCR-VALID representations allows fast and accurate TCR clustering, benchmarked against other state-of-the-art TCR clustering tools and pre-trained language models.

bioinformatics↗

bioRxiv2023

Identification of pan-cancer/testis genes and validation of therapeutic targeting in triple-negative breast cancer: Lin28a- and Siglece-based vaccination induces anti-tumor immunity and inhibits metastasis

BackgroundCancer-testis (CT) genes are targets for tumor antigen-specific immunotherapy given that their expression is normally restricted to the immune-privileged testis in healthy individuals with aberrant expression in tumor tissues. While they represent targetable germ-tissue antigens and play important functional roles in tumorigenesis, there is currently no standardized approach for identifying clinically relevant CT genes. Optimized algorithms and validated methods for accurate prediction of reliable CT antigens with high immunogenicity are also lacking. MethodsSequencing data from the Genotype-Tissue Expression (GTEx) and The Genomic Data Commons (GDC) databases was utilized for the development of a bioinformatic pipeline to identify CT exclusive genes. A CT germness score was calculated based on the number of CT genes expressed within a tumor type and their degree of expression. The impact of tumor germness with clinical outcome was evaluated using healthy GTEx and GDC tumor samples. We then used a triple-negative breast cancer mouse model to develop and test an algorithm that predicts epitope immunogenicity based on the identification of germline sequences with strong MHCI and MHCII binding affinities. Germline sequences for CT genes were synthesized as long synthetic peptide vaccines and tested in the 4T1 triple-negative model of invasive breast cancer with Poly(I:C) adjuvant. Vaccine immunogenicity was determined by flow cytometric analysis of in vitro and in vivo T cell responses. Primary tumor growth and lung metastasis was evaluated by histopathology, flow cytometry and colony formation assay. ResultsWe developed a new bioinformatic pipeline to reliably identify CT exclusive genes as immunogenic targets for immunotherapy. We identified CT genes that are exclusively expressed within the testis, lack detectable thymic expression, and are significantly expressed in multiple tumor types. High tumor germness correlated with tumor progression but not with tumor mutation burden, supporting CT antigens as appealing targets in low mutation burden tumors. Importantly, tumor germness also correlated with markers of anti-tumor immunity. Vaccination of 4T1 tumor bearing mice with Siglece and Lin28a antigens resulted in increased T cell anti-tumor immunity and reduced primary tumor growth and lung metastases. ConclusionOur results present a novel strategy for the identification of highly immunogenic CT antigens for the development of targeted vaccines that induce anti-tumor immunity and inhibit metastasis.

bioinformatics↗

bioRxiv2021

Viral population genomics reveals host and infectivity impact on SARS-CoV-2 adaptive landscape

Public health surveillance, drug treatment development, and optimization of immunological interventions all depend on understanding pathogen adaptation, which differ for specific pathogens. SARS-CoV-2 is an exceptionally successful human pathogen, yet complete understanding of the forces driving its evolution is lacking. Here, we leveraged almost four million SARS-CoV-2 sequences originating mostly from non-vaccinated naive patients to investigate the impact of functional constraints and natural immune pressures on the sequence diversity of the SARS-CoV-2 genome. Overall, we showed that the SARS-CoV-2 genome is under strong and intensifying levels of purifying selection with a minority of sites under diversifying pressure. With a particular focus on the spike protein, we showed that sites under selection were critical for protein stability and virus fitness related to increased infectivity and/or reduced neutralization by convalescent sera. We investigated the genetic diversity of SARS-CoV-2 B and T cell epitopes and determined that the currently known T cell epitope sequences were highly conserved. Outside of the spike protein, we observed that mutations under selection in variants of concern can be associated to beneficial outcomes for the virus. Altogether, the results yielded a comprehensive map of all sites under selection across the entirety of SARS-CoV-2 genome, highlighting targets for future studies to better understand the virus spread, evolution and success.

evolutionary biology↗

bioRxiv2021

In vitro and in vivo preclinical studies predict REGEN-COV protection against emergence of viral escape in humans

Monoclonal antibodies against SARS-CoV-2 are a clinically validated therapeutic option against COVID-19. As rapidly emerging virus mutants are becoming the next major concern in the fight against the global pandemic, it is imperative that these therapeutic treatments provide coverage against circulating variants and do not contribute to development of treatment emergent resistance. To this end, we investigated the sequence diversity of the spike protein and monitored emergence of minor virus variants in SARS-COV-2 isolates found in COVID-19 patients or identified from preclinical in vitro and in vivo studies. This study demonstrates that a combination of non-competing antibodies, REGEN-COV, not only provides full coverage against current variants of concern/interest but also protects against emergence of new such variants and their potential seeding into the population in a clinical setting.

microbiology↗

bioRxiv2020

AdRoit: an accurate and robust method to infer complex transcriptome composition

Bulk RNA sequencing technology provides the opportunity to understand biology at the whole transcriptome level without the prohibitive cost of single cell profiling. Advances in spatial transcriptomics enable to dissect tissue organization and function by genome-wide gene expressions. However, the readout of both technologies is the overall gene expression across potentially many cell types without directly providing the information of cell type constitution. Although several in-silico approaches have been proposed to deconvolute RNA-Seq data composed of multiple cell types, many suffer a deterioration of performance in complex tissues. Here we present AdRoit, an accurate and robust method infer the cell composition from transcriptome data comprised of multiple cell types. AdRoit uses gene expression profile obtained from single cell RNA sequencing as a reference. It employs an adaptive learning approach to correct the sequencing technique difference between the single cell data and the bulk or spatial transcriptome data, enabling cross-platform readout comparability. Our systematic benchmarking and applications, which include deconvoluting complex mixtures that encompass 30 cell types, demonstrate its superior sensitivity and specificity compared to other existing methods as well as its utilities. In addition, AdRoit is computationally efficient and runs orders of magnitude faster than many existing methods.

bioinformatics↗

bioRxiv2020

REGN-COV2 antibody cocktail prevents and treats SARS-CoV-2 infection in rhesus macaques and hamsters

An urgent global quest for effective therapies to prevent and treat COVID-19 disease is ongoing. We previously described REGN-COV2, a cocktail of two potent neutralizing antibodies (REGN10987+REGN10933) targeting non-overlapping epitopes on the SARS-CoV-2 spike protein. In this report, we evaluate the in vivo efficacy of this antibody cocktail in both rhesus macaques and golden hamsters and demonstrate that REGN-COV-2 can greatly reduce virus load in lower and upper airway and decrease virus induced pathological sequalae when administered prophylactically or therapeutically. Our results provide evidence of the therapeutic potential of this antibody cocktail.

microbiology↗

bioRxiv2020

Noise regularization removes correlation artifacts in single-cell RNA-seq data preprocessing

With the rapid advancement of single-cell RNA-seq (scRNA-seq) technology, many data preprocessing methods have been proposed to address numerous systematic errors and technical variabilities inherent in this technology. While these methods have been demonstrated to be effective in recovering individual gene expression, the suitability to the inference of gene-gene associations and subsequent gene networks reconstruction have not been systemically investigated. In this study, we benchmarked five representative scRNA-seq normalization/imputation methods on human cell atlas bone marrow data with respect to their impact on inferred gene-gene associations. Our results suggested that a considerable amount of spurious correlations was introduced during the data preprocessing steps due to over-smoothing of the raw data. We proposed a model-agnostic noise regularization method that can effectively eliminate the correlation artifacts. The noise regularized gene-gene correlations were further used to reconstruct gene co-expression network and successfully revealed several known immune cell modules.

bioinformatics↗

bioRxiv2019

Bayesian inference of power law distributions

Observed data from many research disciplines, ranging from cellular biology to economics, often follow a particular long-tailed distribution known as a power law. Despite the ubiquity of natural power laws, inferring the exact form of the distribution from sampled data remains challenging. The possible presence of multiple generative processes giving rise to an unknown weighted mixture of distinct power law distributions in a single dataset presents additional challenges. We present a probabilistic solution to these issues by developing a Bayesian inference approach, with Markov chain Monte Carlo sampling, to accurately estimate power law exponents, the number of mixtures, and their weights, for both discrete and continuous data. We determine an objective prior distribution that is invariant to reparameterization of parameters, and demonstrate its effectiveness to accurately infer exponents, even in the low sample limit. Finally, we provide a comprehensive and documented software package, written in Python, of our Bayesian inference methodology, freely available at https://github.com/AtwalLab/BayesPowerlaw.

bioinformatics↗

bioRxiv2019

Cell non-autonomous interactions during non-immune stromal progression in the breast tumor microenvironment

SummaryThe breast tumor microenvironment of primary and metastatic sites is a complex milieu of differing cell populations, consisting of tumor cells and the surrounding stroma. Despite recent progress in delineating the immune component of the stroma, the genomic expression landscape of the non-immune stroma (NIS) population and their role in mediating cancer progression and informing effective therapies are not well understood. Here we obtained 52 cell-sorted NIS and epithelial tissue samples across 37 patients from i) normal breast, ii) normal breast adjacent to primary tumor, iii) primary tumor, and iv) metastatic tumor sites. Deep RNA-seq revealed diverging gene expression profiles as the NIS evolves from normal to metastatic tumor tissue, with intra-patient normal-primary variation comparable to inter-patient variation. Significant expression changes between normal and adjacent normal tissue support the notion of a cancer field effect, but extended out to the NIS. Most differentially expressed protein-coding genes and lncRNAs were found to be associated with pattern formation, embryogenesis, and the epithelial-mesenchymal transition. We validated the protein expression changes of a novel candidate gene, C2orf88, by immunohistochemistry staining of representative tissues. Significant mutual information between epithelial ligand and NIS receptor gene expression, across primary and metastatic tissue, suggests a unidirectional model of molecular signaling between the two tissues. Furthermore, survival analyses of 827 luminal breast tumor samples demonstrated the predictive power of the NIS gene expression to inform clinical outcomes. Together, these results highlight the evolution of NIS gene expression in breast tumors and suggest novel therapeutic strategies targeting the microenvironment.

cancer biology↗

bioRxiv2018

T-cell receptor αβ chain pairing is associated with CD4+ and CD8+ lineage specification

While a highly diverse T-cell receptor (TCR) repertoire is the hallmark of a healthy adaptive immune system, relatively little is understood about how the CD4+ and CD8+ TCR repertoires differ from one another. We here utilize high-throughput single T-cell sequencing to obtain approximately 100,000 TCR {beta} chain pairs from human subjects, stratified into CD4+ and CD8+ lineages. We reveal that substantial information about T-cell lineage is encoded by V{beta} gene pairs and, to a lesser extent, by several other TCR features such as CDR3 length and charge. We further find that the strength of association between the {beta} chain and T-cell lineage is surprisingly weak, similar in strength to that of the chain. Using machine learning classifiers to predict T-cell lineage from TCR features, we demon-strate that {beta} chain pairs are significantly more informative than individual chains alone. These findings provide unprecedented insight into the CD4+ and CD8+ TCR repertoires and highlight the importance of {beta} chain pairing in TCR function and specificity.

immunology↗

bioRxiv2017

Single-cell sequencing reveals αβ chain pairing shapes the T cell repertoire

A diverse T cell repertoire is a critical component of the adaptive immune system, providing protection against invading pathogens and neoplastic changes, relying on the recognition of foreign antigens and neoantigen peptides by T cell receptors (TCRs). However, the statistical properties and function of the T cell pool in an individual, under normal physiological conditions, are poorly understood. In this study, we report a comprehensive, quantitative characterization of the T cell repertoire from over 1.9 million cells, yielding over 200,000 high quality paired {beta} sequences in 5 healthy human subjects. The dataset was obtained by leveraging recent biotechnology developments in deep RNA sequencing of lymphocytes via single-cell barcoding in emulsion. We report non-random associations and non-monogamous pairing between the and {beta} chains, lowering the theoretical diversity of the T cell repertoire, and increasing the frequency of public clones shared among individuals. T cell clone size distributions closely followed a power law, with markedly longer tails for CD8+ cytotoxic T cells than CD4+ helper T cells. Furthermore, clonality estimates based on paired chains from single T cells were lower than that from single chain data. Taken together, these results highlight the importance of sequencing {beta} pairs to accurately quantify lymphocyte receptor diversity.

immunology↗