Biology subjects

Islam, S. M. S.

Publications and source records attributed to Islam, S. M. S..

5 recordsLinked to original sources

bioRxiv2026

Sum-h2, enabling genetic discovery for deep learning-derived phenotypes through a fast evaluation framework and arena of performance

In the recent growing interest of AI research toward biology, genetic association studies of AI- derived phenotypes from high-content modalities such as images emerges as a powerful means for biological discovery. However, such AI-phenotyping methods still lacks a good optimization target and an efficient evaluation framework. The number of discovered loci was the major criterion for evaluating the quality of AI-derived phenotypes. However, the time and computational resources required for running the GWAS and subsequent loci-clumping are substantial, limiting the rapid development and iteration of deep learning representation algorithms. Here we present a 1000x faster and lightweight framework, sum-h2, than traditional GWAS framework for evaluating genetic discovery through total heritability. We revisit tr(P^(-1) G), previously proposed in the context of evolutionary studies, as a measure of multi-phenotype heritability. We showed that sum-h2, the sum of heritability over phenotypic Principal Components (PCs), is equivalent to the linear transformation-invariant tr(P^(-1) G), through both theoretical proof and simulation studies. Moreover, sum-h2 can be estimated rapidly with minimal information loss over a relatedness-enriched sample, while preserving the relative ranking of endophenotypes by GWAS loci counts. Based on selected UKB data and sum-h2, we set up a Genetic Discovery Arena, enabling rapid and fair comparisons for the development and optimization of deep learning-derived phenotyping methods.

bioinformatics↗

bioRxiv2025

HIV-Induced Sialoglycans on Infected Cells Promote Immune Evasion from Myeloid Cell-Mediated Killing

Sialic acid-containing glycans (sialoglycans) on pathological cells interact with Siglecs, glyco-immune checkpoint receptors expressed on myeloid cells such as monocytes and neutrophils. This interaction suppresses the cytotoxic functions of these immune cells. We show that HIV infection reprograms the glycosylation machinery of infected cells to increase the expression of specific sialoglycan ligands for Siglecs-3, -7, and -9. These ligands engage Siglecs on myeloid cells, impairing their ability to target HIV-infected cells. Selective disruption of these interactions using 10-1074-Sia, an HIV-specific antibody conjugated to sialidase--an enzyme that removes sialic acids--significantly enhances monocyte- and neutrophil-mediated killing of HIV-infected cells in autologous assays. Treatment with 10-1074-Sia in humanized mice infected with HIV reduces viral load and decreases inflammation. These findings reveal a novel immune evasion mechanism exploited by HIV to evade myeloid cell immune surveillance and highlight the potential of targeting sialoglycan-Siglec interactions to improve immune clearance of HIV-infected cells.

immunology↗

bioRxiv2024

NeuroAtlas: An Artificial Intelligence-based Framework for Annotation, Segmentation and Registration of Large Scale Biomedical Imaging Data

With increasing neuroimaging modalities and data diversity, mapping brain regions to a standard atlas template has become a challenging problem. Machine learning in general and deep learning, in particular, have been providing robust solutions for several neuroimaging tasks, including brain image registration and segmentation. However, these methods require a large amount of data for groundtruth labels, annotated by human experts, which is time-consuming. In this work, we introduce NeuroAtlas, an AI-based framework for atlas generation and brain region segmentation. We showcase an end-to-end solution for brain registration and segmentation by providing i) a deep learning modeling suite with a variety of high-performing model architectures to map a brain atlas onto the input brain section and ii) a Graphical User Interface (GUI)-based plugin for large-scale data annotation with a feature of modifying the predicted labels for active learning. We demonstrate a robust performance of our framework on the human brains, captured through various imaging modalities and age groups, and demonstrate its application for mouse brains as well. NeuroAtlas tool will be open-sourced and entirely compatible with both local as well as cloud-based computing so that users can easily adapt to their neuroimaging custom datasets.

neuroscience↗

bioRxiv2024

Multimodal 3D Image Registration for Mapping Brain Disorders

We introduce an AI-driven approach for robust 3D brain image registration, addressing challenges posed by diverse hardware scanners and imaging sites. Our model trained using an SSIM-driven loss function, prioritizes structural coherence over voxel-wise intensity matching, making it uniquely robust to inter-scanner and intra-modality variations. This innovative end-to-end framework combines global alignment and non-rigid registration modules, specifically designed to handle structural, intensity, and domain variances in 3D brain imaging data. Our approach outperforms the baseline model in handling these shifts, achieving results that align closely with clinical ground-truth measurements. We demonstrate its efficacy on 3D brain data from healthy individuals and dementia patients, with particular success in quantifying brain atrophy, a key biomarker for Alzheimers disease and other brain disorders. By effectively managing variability in multisite, multi-scanner neuroimaging studies, our approach enhances the precision of atrophy measurements for clinical trials and longitudinal studies. This advancement promises to improve diagnostic and prognostic capabilities for neurodegenerative disorders.

neuroscience↗

bioRxiv2018

Identification of Ovarian Cancer Gene Expression Patterns Associated with Disease Progression and Mortality

Ovarian cancer (OC) is a common cause of death from cancer among women worldwide, so there is a pressing need to identify factors influencing mortality. Much OC patient clinical data is now publically accessible (including patient age, cancer site stage and subtype), as are large datasets of OC gene transcription profiles. These have enabled studies correlating OC patient survival with clinical variables and with gene expression but it is not well understood how these two aspects interact to influence mortality. To study this we integrated clinical and tissue transcriptome data from the same patients available from the Broad Institute Cancer Genome Atlas (TCGA) portal. We investigated OC mRNA expression levels (relative to normal patient tissue) of 26 genes already strongly implicated in OC, assessed how their expression in OC tissue predicts patient survival then employed Cox Proportional Hazard regression models to analyse both clinical factors and transcriptomic information to determine relative risk of death associated with each factor. Multivariate analysis of combined data (clinical and gene mRNA expression) found age, ovary tumour site and cancer stage IB significantly correlated with patient survival. Univariate analysis also confirmed significant differences in patient survival time when altered transcription levels of KLK6, CD36, MEF2C and SCGB2A1 were evident, while multivariate analysis that considered the 26 genes simultaneously revealed a significant relationship of mortality with KLK6, CD36 and E2F1 genes. However, analysis that considered all 26 genes with clinical variables together identified WFDC2, E2F1, BRCA1, KLK6, SCGB2A1 and SLPI genes as independently related to mortality in OC. This indicated that the latter genes affect OC patient survival, i.e., provided mechanistic and predictive information in addition to that of the clinical traits and provide strong evidence that these genes are critical markers of processes that underlie OC progression and mortality.

bioinformatics↗