Biology subjects

Shirali, A.

Publications and source records attributed to Shirali, A..

7 recordsLinked to original sources

bioRxiv2025

Conditional Protein Structure Generation with Protpardelle-1c

We present Protpardelle-1c, a collection of protein structure generative models with robust motif scaffolding and support for multi-chain complex generation under hotspot-conditioning. Enabling sidechain-conditioning to a backbone-only model increased Protpardelle-1cs MotifBench score from 4.97 to 28.16, outperforming RFdiffusions 21.27. The crop-conditional all-atom model achieved 208 unique solutions on the La-Proteina all-atom motif scaffolding benchmark, on par with La-Proteina while having ~10 times fewer parameters. At 22M parameters, Protpardelle-1c enables rapid sampling, taking 40 minutes to sample all 3000 MotifBench backbones on an NVIDIA A100-80GB, compared to 31 hours for RFdiffusion.

bioengineering↗

bioRxiv2025

Homology-Based Enzymatic Assembly of Modular T7 Phage Genome

Bacteriophages, viruses that infect bacteria, are pivotal in therapeutic, industrial, and bio-detection applications due to their unique ability to inject DNA into bacterial hosts and change the genetics and behavior of whole bacterial populations. Genetic engineering of phage genomes has expanded their potential applications with several established methods such as Golden Gate assembly, yeast cloning, {lambda} Red recombineering, and CRISPR-Cas systems. Here, we present a novel, efficient method to design and make synthetic bacteriophages in vitro without using restriction enzymes to allow for modular insertion of DNA fragments into the phage genome. The ability to create synthetic bacteriophages in vitro without the use of restriction enzymes allows for simpler engineering without the hassle or cloning limitations encountered when building domesticated bacteriophage genomes.

bioengineering↗

bioRxiv2023

A Comprehensive Survey of Scoring Functions for Protein Docking Models

The protein-protein docking problem is fundamental to our understanding of how proteins interact. It is also important because of its relevance to drug discovery and vaccine design. Docking methods typically involve two steps. The first step generates many possible candidate conformations, while the second step evaluates the candidates using some scoring function. Despite numerous experiments with many innovative scoring functions, a good scoring function for docking remains elusive. Deep learning models offer alternatives to using explicit empirical or mathematical functions for scoring protein-protein complexes. In this study, we perform a comprehensive survey of the state-of-the-art scoring functions by considering the most popular and highly performant approaches for scoring protein-protein complexes, evaluating their strengths and weaknesses to aid researchers in understanding progress made in this field. Our survey compares the state-of-the-art in classical and deep learning-based scoring methods.

molecular biology↗

bioRxiv2023

DNA-targeting and cell-penetrating antibody-drug conjugate

DNA released by dying cancer cells offers a tumor targeting strategy that is independent of specific cell surface antigens. Anti-DNA antibodies preferentially localize to tumor microenvironments enriched in extracellular DNA and can penetrate live tumor cells through nucleoside salvage pathways. Nuclear-localizing variants of anti-DNA antibodies cause DNA damage and selectively kill cancer cells with defects in DNA repair. Here we show that an optimized full-length IgG1 anti-DNA antibody penetrates live cells and is synthetically lethal to BRCA2-deficient tumors but has minimal effect of DNA repair-proficient tumors. Linkage of the antibody to the anti-mitotic drug monomethyl auristatin E yields a DNA-targeting and cell-penetrating anti-DNA antibody-drug conjugate (ADC) that is well tolerated in mice and highly toxic to tumors with intact DNA repair. This work provides proof-of-concept for the novel use of an anti-DNA antibody as the backbone of a DNA-targeting, cell-penetrating ADC that can impact tumors that otherwise lack specifically targetable surface antigens. Statement of significanceA strategy for targeting tumors that lack specific surface antigens is revealed by an anti-DNA antibody-drug conjugate that localizes to tumor microenvironments enriched in DNA and penetrates cells through nucleoside salvage pathways.

immunology↗

bioRxiv2023

cGAS-activating lupus autoantibody for cancer immunotherapy

Cytoplasmic DNA triggers a cGAS-mediated signaling cascade that promotes an innate immune response and is potentially actionable in cancer immunotherapy. Here we show that a cytoplasmic-localizing lupus anti-DNA autoantibody activates cGAS and facilitates an immune-mediated prolongation of survival in orthotopic models of glioblastoma (GBM). Mechanistically, cellular penetration and blood-brain barrier crossing by the anti-DNA autoantibody is linked to nucleoside transport. Pulldown, knockdown, signaling, and cytotoxicity assays demonstrate autoantibody association with and activation of cGAS. In orthotopic GBM models, the autoantibody localizes to brain tumor, increases tumor CD8+ T cell content, and prolongs survival in immunocompetent but not immunodeficient mice. This work introduces the new concept of a cGAS-activating anti-DNA autoantibody, which impacts theories on mechanisms of autoimmunity and has translational applications in cancer immunotherapy.

immunology↗

bioRxiv2023

PIsToN: Evaluating Protein Binding Interfaces with Transformer Networks

The computational studies of protein binding are widely used to investigate fundamental biological processes and facilitate the development of modern drugs, vaccines, and therapeutics. Scoring functions aim to predict complexes that would be formed by the binding of two biomolecules and to assess and rank the strength of the binding at the interface. Despite past efforts, the accurate prediction and scoring of protein binding interfaces remain a challenge. The physics-based methods are computationally intensive and often have to trade accuracy for computational cost. The possible limitations of current machine learning (ML) methods are ineffective data representation, network architectures, and limited training data. Here, we propose a novel approach called PIsToN (evaluating Protein binding Interfaces with Transformer Networks) that aim to distinguish native-like protein complexes from decoys. Each protein interface is transformed into a collection of 2D images (interface maps), where each image corresponds to a geometric or biochemical property in which pixel intensity represents the feature values. Such a data representation provides atomic-level resolution of relevant protein characteristics. To build hybrid machine learning models, additional empirical-based energy terms are computed and provided as inputs to the neural network. The model is trained on thousands of native and computationally-predicted protein complexes that contain challenging examples. The multi-attention transformer network is also endowed with explainability by highlighting the specific features and binding sites that were the most important for the classification decision. The developed PIsToN model significantly outperforms existing state-of-the-art scoring functions on well-known datasets.

bioinformatics↗

bioRxiv2020

Using distance on the Riemannian manifold to compare representations in brain and models

Representational similarity analysis (RSA) summarizes activity patterns for a set of experimental conditions into a matrix composed of pairwise comparisons between activity patterns. Two examples of such matrices are the condition-by-condition inner product matrix or the correlation matrix. These representational matrices reside on the manifold of positive semidefinite matrices, called the Riemannian manifold. We hypothesize that representational similarities would be more accurately quantified by considering the underlying manifold of the representational matrices. Thus, we introduce the distance on the Riemannian manifold as a metric for comparing representations. Analyzing simulated and real fMRI data and considering a wide range of metrics, we show that the Riemannian distance is least susceptible to sampling bias, results in larger intra-subject reliability, and affords searchlight mapping with high sensitivity and specificity. Furthermore, we show that the Riemannian distance can be used for measuring multi-dimensional connectivity. This measure captures both univariate and multivariate connectivity and is also more sensitive to nonlinear regional interactions compared to the state-of-the-art measures. Applying our proposed metric to neural network representations of natural images, we demonstrate that it also possesses outstanding performance in quantifying similarity in models. Taken together, our results lend credence to the proposition that RSA should consider the manifold of the representational matrices to summarize response patterns in the brain and models.

neuroscience↗