Biology subjects

Gholami, S.

Publications and source records attributed to Gholami, S..

7 recordsLinked to original sources

bioRxiv2025

The TEAIV model: Extending the standard TEIV model to account for viral budding ramp up

Understanding how viruses replicate and spread within a host is fundamental to predicting disease progression, timing, and dosage of effective therapeutic and prophylactic interventions. A target-cell limited approach is often used to model within-host viral kinetics to characterize disease infection dynamics. The standard target-cell limited model, the TEIV model, has been instrumental for understanding SARS-CoV-2 within-host viral kinetics, however, its core assumptions of instantaneous viral budding and exponentially distributed cell lifetimes oversimplify fundamental biological processes, potentially limiting predictive power. In this work, we consider a novel model approach, the TEAIV model: an extension to the TEIV model that considers an infected partially-productive cell state that ramps up through n stages to a fully productive state. We fit the TEAIV model to various SARS-CoV-2 viral load datasets, separated by hospitalized (severe infection) and non-hospitalized (mild infection) individual data, consider multiple different ramping functions, and compare to standard TEIV model fits. We further compare TEAIV and TEIV model fits with and without infected cell death incorporated in the eclipse (E) and partially infected (A) compartments and investigate best-fit frameworks up to 10 A states. We find that linear budding ramp-up dynamics minimizes the BIC (with {Delta}BIC > 2) across all model formulations when the total number of eclipse and budding compartments exceeds three. Furthermore, we find that the inclusion or exclusion of cell death applied to eclipse or ramping compartments does not substantially affect this result. Finally, we analytically consider the most general TEAIV model and discuss future modelling considerations. Our results demonstrate that accounting for non-instantaneous viral budding provides a better fit to SARS-CoV-2 viral load data and establishes a foundation for more mechanistically informed within-host models.

systems biology↗

bioRxiv2025

Mathematical Modeling of Bone Remodeling after Surgical Menopause

Osteoporosis is a skeletal pathology characterized by decreased bone mass and structural deterioration resulting from an imbalance in bone metabolic processes. Estrogen deficiency in postmenopausal women leads to an increased risk of osteoporosis, while women who have undergone complete oophorectomies display an even higher risk due to the sudden decrease in estrogen. Some evidence indicates that bone loss slows in the period beyond 15 years after surgery; however, there is substantial uncertainty in clinical data. To explore the effects of surgically induced menopausal transition, here we propose a mathematical model for the bone cell dynamical responses to sudden estrogen deficiency, which extends an existing model for osteoporosis due to aging and natural menopause. Using data on the key impacts observed in female mice and humans after bilateral oophorectomy, this new model considers the role of osteocytes embedded within the bone mineralized matrix in regulating osteoclastogenesis, which results in increased bone resorption after surgical menopause. The values of model parameters in natural and surgical menopause were estimated from aggregated human clinical data from existing longitudinal studies. The new model effectively captures the previously unmodeled increase in bone loss during the first 15 years post-surgical menopause and the rebound in bone mineral density in the long-term. With this model, effects of treatments on targeting osteocyte dynamics could be explored in the future.

systems biology↗

bioRxiv2025

Compact Vision--Language Models Enable Efficient and Interpretable Automated OCT Analysis Through Layer Specific Multimodal Learning

Translating the intricate anatomical signatures of retinal disease from OCT B-scans into clear, accurate clinical narratives demands AI models that seamlessly fuse visual features with domain expertise. We curated a multimodal dataset of 40,000 OCT B-scans from public repositories and private clinical cohorts, each paired with expert validated summaries spanning six conditions: diabetic macular edema, diabetic retinopathy, geographic atrophy, drusen, choroidal neovascularization, and healthy retina. We introduce LO-VLM, a compact (247M parameter) vision-language model (VLM) that infuses anatomical guidance into both encoder and decoder for free form summary generation and multiclass disease classification. Benchmarking against state-of-the-art RetinaVLM, LLaVA-Med, and a ViT vision only model demonstrates superior performance. In a blinded evaluation by three board certified retina specialists scored the generated summaries, LO-VLM narratives achieved mean = 8.5 (standard deviation = 1.15) out of 10, compared to mean = 5.5 (standard deviation = 1.13) for RetinaVLM (p < 0.0001). In quantitative evaluations, LO-VLM achieved an SBERT similarity of 0.803 and a BERTScore F1 of 0.715, representing improvements of 8.2% and 28.8% over specialized VLM baselines. For disease classification, LO-VLM reached 96% accuracy (F1 = 96%), outperforming ViT by 13% and exceeding medical VLM benchmarks by over 62%. By reconciling interpretability with computational efficiency, LO-VLM establishes a new paradigm for efficient AI models in OCT interpretation.

bioengineering↗

bioRxiv2025

Targeting Neutrophil Extracellular Traps to inhibit Colon Cancer Tumor Necrosis and Metastasis

Necrosis, conventionally thought of as a passive consequence of aggressive tumor growth, is associated with poor prognosis in colorectal cancer (CRC). We recently discovered that necrosis can be caused by neutrophils and neutrophil extracellular traps (NETs) aggregates driving vascular occlusion within the tumor vasculature in models of breast cancer. Here, we evaluated the role of NETs in inducing necrosis and metastasis in CRC. We found that the numbers of neutrophils primed to form NETs were elevated in the circulation of patients with CRC as compared to controls. CD177Low neutrophils were also elevated, and they showed reduced extravasation capacity with intact ability to form NETs. The extent of necrosis correlated with metastasis (stage IV disease), independent of tumor size, in our human cohort. In both human and murine CRC tumors, necrotic regions were characterized by neutrophil infiltration and NET accumulation, and NET aggregates were observed in the vasculature next to the necrotic regions. Single cell RNA sequencing and spatial transcriptomic analysis of human CRC and liver metastases revealed that necrotic tumors activate pathways associated with increased metastatic potential, including epithelial-to-mesenchymal-transition. Using a mouse model of DNA mismatch repair proficient CRC, we found neutrophil infiltration and NETs increased with tumor progression. Genetic or pharmacological inhibition of NET formation decreased necrosis and metastasis, and importantly enhanced chemotherapy efficacy. Altogether, our findings show that NET formation in human CRC is a key feature of tumor necrosis, that it is associated with metastasis, and further suggest that preventing NET formation may offer clinical benefits to CRC patients.

cancer biology↗

bioRxiv2024

EYE-Llama, an in-domain large language model for ophthalmology

1BackgroundTraining Large Language Models (LLMs) with in-domain data can significantly enhance their performance, leading to more accurate and reliable question-answering (Q&A) systems essential for supporting clinical decision-making and educating patients. MethodsThis study introduces ophthalmic LLMs trained on in-domain, well-curated datasets. We present an open-source substantial ophthalmic language dataset for model training. Our models (EYE-Llama), were pre-trained on an ophthalmology-specific dataset, including paper abstracts, textbooks, and Wikipedia articles. Subsequently, the models underwent fine-tuning using a diverse range of QA pairs. Our models were compared to baseline Llama 2, ChatDoctor, Meditron, Llama 3, and ChatGPT (GPT3{middle dot}5) models, using four distinct test sets, and evaluated quantitatively (Accuracy, F1 score, BERTScore, BARTScore and BLEU score) and qualitatively by two ophthalmologists. FindingsUpon evaluating the models using the synthetic dialogue test set with three different metrics (BERTScore, BARTScore, and BLEU score), our models demonstrated superior performance. Specifically, when evaluated using BERTScore, our models surpassed Llama 2, Llama 3, Meditron, and ChatDoctor in terms of F1 score, and performed on par with ChatGPT, which was trained with 175 billion parameters (EYE-Llama: 0.57, Llama 2: 0.56, Llama 3: 0.55, Meditron: 0.50, ChatDoctor: 0.56, and ChatGPT: 0.57). Additionally, the EYE-Llama model outperformed the above models when evaluated using BARTScore and BLEU scores. When tested on the MedMCQA test set, the fine-tuned models exhibited higher accuracy compared to Llama 2, Meditron, and ChatDoctor models (EYE-Llama: 0.39, Llama 2: 0.33, ChatDoctor: 0.29, Meditron: 0.22). However, ChatGPT, and Llama 3 models outperformed EYE-Llama, achieving accuracies of 0.55, 0.78, and 0.90, respectively. On the PubmedQA test set, our model showed improved accuracy over all other models (EYE-Llama: 0.96, Llama 2: 0.90, Llama 3: 0.92, Meditron: 0.76, ChatGPT: 0.93, ChatDoctor: 0.92). InterpretationThe study shows that pre-training and fine-tuning LLMs like EYE-Llama enhances their performance in specific medical domains. Our EYE-Llama models surpass baseline Llama 2 in all evaluations, highlighting the effectiveness of specialized LLMs in medical QA systems. FundingFunded by NEI R15EY035804 (MNA), R21EY035271 (MNA), and UNC Charlotte Faculty Research Grant (MNA)

bioengineering↗

bioRxiv2024

Transport mechanism of DgoT, a bacterial homolog of SLC17 organic anion transporters

The solute carrier 17 (SLC17) family contains anion transporters that accumulate neuro-transmitters in secretory vesicles, remove carboxylated monosaccharides from lysosomes, or extrude organic anions from the kidneys and the liver. We combined classical molecular dynamics simulations, Markov state modeling and hybrid first principles quantum mechani-cal/classical mechanical (QM/MM) simulations with experimental approaches to describe the transport mechanisms of a model bacterial protein, the D-galactonate transporter DgoT, at atomic resolution. We found that protonation of D46 and E133 precedes galactonate binding and that substrate binding induces closure of the extracellular gate, with the conserved R47 coupling substrate binding to transmembrane helix movement. After isomerization to an inward-facing conformation, deprotonation of E133 and subsequent proton transfer from D46 to E133 opens the intracellular gate and permits galactonate dissociation either in its unprotonated form or after proton transfer from E133. After release of the second proton, apo DgoT returns to the outward-facing conformation. Our results provide a framework to understand how various SLC17 transport functions with distinct transport stoichiometries can be attained through subtle variations in proton and substrate binding/unbinding.

biophysics↗

bioRxiv2023

Federated learning for diagnosis of age-related macular degeneration

This paper presents a federated learning (FL) approach to train deep learning models for classifying age-related macular degeneration (AMD) using optical coherence tomography image data. We employ the use of residual networks and vision transformer encoders for the normal vs AMD binary classification, integrating four unique domain adaptation techniques to address domain shift issues caused by heterogeneous data distribution in different institutions. Experimental results indicate that FL strategies can achieve competitive performance similar to centralized models even though each local model has access to a portion of the training data. Notably, Adaptive Personalization FL strategy stood out in our FL evaluations, consistently delivering high performance across all tests due to its additional local model. Furthermore, the study provides valuable insights into the efficacy of simpler architectures in image classification tasks, particularly in scenarios where data privacy and decentralization are critical using both encoders. It suggests future exploration into deeper models and other FL strategies for a more nuanced understanding of these models performance.

bioengineering↗