Biology subjects

Muhammed, L.

Publications and source records attributed to Muhammed, L..

4 recordsLinked to original sources

bioRxiv2025

TP53 and RB1 are predictive genetic biomarkers for sensitivity to cytarabine in gliomas

Therapeutic progress in glioma, one of the most lethal human cancers, has been limited by molecular heterogeneity and lack of biomarker-driven drug deployment. Here we used a proprietary large-scale CRISPRi screening in primary patient-derived glioma tumorspheres to identify genetic vulnerabilities and nominate pharmacologically tractable targets. DNA polymerase-linked dependencies emerged as top-ranked hits, which we validated through orthogonal viability assays. Network-based integration of dependency data with drug-target relationships nominated cytarabine, a nucleoside analogue already approved for intrathecal use, as a candidate agent targeting this axis. Dose-response profiling across molecularly diverse glioma models revealed substantial heterogeneity in cytarabine sensitivity (IC50 range: 0.04-9.8 {micro}M). Machine learning analysis of whole-genome sequencing data identified TP53 wild-type and RB1-wild-type status as dominant predictors of response, with double wild-type lines showing three standard deviations (3 s.d.) increased sensitivity compared to altered models. Prospective validation in an independent cohort confirmed that TP53/RB1 genotype stratifies cytarabine activity. These findings establish a mechanistically anchored, biomarker-restricted repurposing opportunity for cytarabine in leptomeningeal glioma, enabling rational prioritisation of an accessible therapy in a molecularly defined patient subset.

cancer biology↗

bioRxiv2025

Multi-modal single-cell foundation models via dynamic token adaptation

AO_SCPLOWBSTRACTC_SCPLOWRecent advances in applying deep learning in genomics include DNA-language and single-cell foundation models. However, these models take only one data type as input. We introduce dynamic token adaptation and demonstrate how it combines these models to predict gene regulation at the single-cell level in different genetic contexts. Although the method is generalisable, we focus on an illustrative example by training an adapter from DNA-sequence embeddings to a single-cell foundation models token embedding space. As a qualitative evaluation, we assess the impact of DNA sequence changes on the models learned gene regulatory networks by mutating the transcriptional start site of the transcription factor GATA4 in silico, observing predicted expression changes in its target genes in fetal cardiomyocytes.

bioinformatics↗

bioRxiv2025

Detecting cell-level transcriptomic changes of Perturb-seq using Contrastive Fine-tuning of Single-Cell Foundation Models

Genome-scale perturbation cell atlases are an exciting new resource to understand the transcriptomic and phenotypic impact of single-gene activation or knockdown. However, in terms of differentially expressed genes identified, the signal detected in these data atlases is low, leading to the exclusion of most data from downstream analyses. Recent advances in single-cell foundation models have shown promise in capturing complex biological insights. However, their application to perturbation analysis, especially in predicting perturbed single-cell transcriptomes, remains limited. In this paper, we focus on learning representations of single-cell transcriptomes that capture subtle, yet important, transcriptome-wide changes, and we propose a novel fine-tuning strategy using contrastive learning to leverage single-cell foundation models for this task. We pre-train a single-cell foundation model and fine-tune on a genome-scale perturbation dataset using a contrastive loss, which minimises the distance between cell embeddings from unperturbed cells while maximising the distance between perturbed and unperturbed cells. We validate and test the model on unseen perturbations, demonstrating its ability to identify global biologically meaningful transcriptional changes not captured by traditional differential expression methods. Our approach provides a novel framework for analysing single-cell perturbation data and offers a more effective means of identifying perturbations that drive systemic gene expression changes.

bioinformatics↗

bioRxiv2022

Single-cell Mendelian randomisation identifies cell-type specific genetic effects on human brain disease and behaviour

Translating genome-wide association loci to therapies requires knowledge of the causal genes, their directionality of effect and the cell-types in which they act. To infer these relationships in the human brain, we implemented Mendelian randomisation using single cell-type expression quantitative trait loci (eQTLs) as genetic anchors. Expression QTLs were mapped across 8 major cell-types in brain tissue exclusively ascertained from donors with no history of brain disease. We report evidence for a causal association between the change in expression of 118 genes and one or more of 16 brain phenotypes, revealing candidate targets for risk mitigation and opportunities for shared preventative therapeutic strategies. We highlight key causal genes for neurodegenerative and neuropsychiatric disease and for each, we report its cellular context and the therapeutic directionality required for risk mitigation. Our use of control samples establishes a new resource for the causal interpretation of GWAS risk alleles for human brain phenotypes.

neuroscience↗