Biology subjects

Popescu, M.

Publications and source records attributed to Popescu, M..

5 recordsLinked to original sources

bioRxiv2024

A Comprehensive Evaluation of Large Language Models in Mining Gene Interactions and Pathway Knowledge

BackgroundUnderstanding complex biological pathways, including gene-gene interactions and gene regulatory networks, is critical for exploring disease mechanisms and drug development. Manual literature curation of biological pathways is useful but cannot keep up with the exponential growth of the literature. Large-scale language models (LLMs), notable for their vast parameter sizes and comprehensive training on extensive text corpora, have great potential in automated text mining of biological pathways. MethodThis study assesses the effectiveness of 21 LLMs, including both API-based models and open-source models. The evaluation focused on two key aspects: gene regulatory relations (specifically, activation, inhibition, and phosphorylation) and KEGG pathway component recognition. The performance of these models was analyzed using statistical metrics such as precision, recall, F1 scores, and the Jaccard similarity index. ResultsOur results indicated a significant disparity in model performance. Among the API-based models, ChatGPT-4 and Claude-Pro showed superior performance, with an F1 score of 0.4448 and 0.4386 for the gene regulatory relation prediction, and a Jaccard similarity index of 0.2778 and 0.2657 for the KEGG pathway prediction, respectively. Open-source models lagged their API-based counterparts, where Falcon-180b-chat and llama1-7b led with the highest performance in gene regulatory relations (F1 of 0.2787 and 0.1923, respectively) and KEGG pathway recognition (Jaccard similarity index of 0.2237 and 0. 2207, respectively). ConclusionLLMs are valuable in biomedical research, especially in gene network analysis and pathway mapping. However, their effectiveness varies, necessitating careful model selection. This work also provided a case study and insight into using LLMs as knowledge graphs.

bioinformatics↗

bioRxiv2023

Iterative Prompt Refinement for Mining Gene Relationships from ChatGPT

ChatGPT has demonstrated its potential as a surrogate knowledge graph. Trained on extensive data sources, including open-access publications, peer-reviewed research articles and biomedical websites, ChatGPT extracted information on gene relationships and biological pathways. However, a major challenge is model hallucination, i.e., high false positive rates. To assess and address this challenge, we systematically evaluated ChatGPTs capacity for predicting gene relationships using GPT-3.5-turbo and GPT-4. Benchmarking against the KEGG Pathway Database as the ground truth, we experimented with diverse prompting strategies, targeting gene relationships of activation, inhibition, and phosphorylation. We introduced an innovative iterative prompt refinement technique. By assessing prompt efficacy using metrics like F-1 score, precision, and recall, GPT-4 was re-engaged to suggest improved prompts. A refined prompt, which combines a specialized role with explanatory text, significantly enhances the performance. Going beyond pairwise gene relationships, we also deciphered complex gene interplays, such as gene interaction chains and pathways pertinent to diseases like non-small cell lung cancer. Direct prompts showed limited success, but "least-to-most" prompting exhibited significant potentials for such network constructions. The methods in this study may be used for some other bioinformatics prediction problems.

bioinformatics↗

bioRxiv2023

pathCLIP: Detection of Genes and Gene Relations from Biological Pathway Figures through Image-Text Contrastive Learning

In biomedical literature, biological pathways are commonly described through a combination of images and text. These pathways contain valuable information, including genes and their relationships, which provide insight into biological mechanisms and precision medicine. Curating pathway information across the literature enables the integration of this information to build a comprehensive knowledge base. While some studies have extracted pathway information from images and text independently, they often overlook the correspondence between the two modalities. In this paper, we present a pathway figure curation system named pathCLIP for identifying genes and gene relations from pathway figures. Our key innovation is the use of an image-text contrastive learning model to learn coordinated embeddings of image snippets and text descriptions of genes and gene relations, thereby improving curation. Our validation results, using pathway figures from PubMed, showed that our multimodal model outperforms models using only a single modality. Additionally, our system effectively curates genes and gene relations from multiple literature sources. A case study on extracting pathway information from non-small cell lung cancer literature further demonstrates the usefulness of our curated pathway information in enhancing related pathways in the KEGG database.

pathology↗

bioRxiv2023

Monocytes use protrusive forces to generate migration paths in viscoelastic collagen-based extracellular matrices

Circulating monocytes are recruited to the tumor microenvironment, where they can differentiate into macrophages that mediate tumor progression. To reach the tumor microenvironment, monocytes must first extravasate and migrate through the type-1 collagen rich stromal matrix. The viscoelastic stromal matrix around tumors not only stiffens relative to normal stromal matrix, but often exhibits enhanced viscous characteristics, as indicated by a higher loss tangent or faster stress relaxation rate. Here, we studied how changes in matrix stiffness and viscoelasticity, impact the three-dimensional migration of monocytes through stromal-like matrices. Interpenetrating networks of type-1 collagen and alginate, which enable independent tunability of stiffness and stress relaxation over physiologically relevant ranges, were used as confining matrices for three-dimensional culture of monocytes. Increased stiffness and faster stress relaxation independently enhanced the 3D migration of monocytes. Migrating monocytes have an ellipsoidal or rounded wedge-like morphology, reminiscent of amoeboid migration, with accumulation of actin at the trailing edge. Matrix adhesions were dispensable for monocyte migration in 3D, but migration did require actin polymerization and myosin contractility. Mechanistic studies indicate that actin polymerization at the leading edge generates protrusive forces that open a path for the monocytes to migrate through in the confining viscoelastic matrices. Taken together, our findings implicate matrix stiffness and stress relaxation as key mediators of monocyte migration and reveal how monocytes use pushing forces at the leading edge mediated by actin polymerization to generate migration paths in confining viscoelastic matrices. Significance StatementCell migration is essential for numerous biological processes in health and disease, including for immune cell trafficking. Monocyte immune cells migrate through extracellular matrix to the tumor microenvironment where they can play a role in regulating cancer progression. Increased extracellular matrix (ECM) stiffness and viscoelasticity have been implicated in cancer progression, but the impact of these changes in the ECM on monocyte migration remains unknown. Here, we find that increased ECM stiffness and viscoelasticity promote monocyte migration. Interestingly, we reveal a previously undescribed adhesion-independent mode of migration whereby monocytes generate a path to migrate through pushing forces at the leading edge. These findings help elucidate how changes in the tumor microenvironment impact monocyte trafficking and thereby disease progression.

cell biology↗

bioRxiv2020

Filamentous Bacteriophage Delay Healing of Pseudomonas-Infected Wounds

We have identified a novel role for filamentous bacteriophage in the delayed healing associated with chronic Pseudomonas aeruginosa (Pa) wound infections. Pf phage delays wound re-epithelialization in the absence of live Pa, indicative that Pf effects on wound healing are independent of Pa pathogenesis. Pf phage directly inhibits autocrine signaling of CXCL1 (KC) to impede keratinocyte migration and wound re-epithelization. In agreement with these studies, a prospective cohort study of 36 human patients with chronic Pa wound infections revealed that wounds infected with Pf positive strains of Pa took longer to heal and were more likely to increase in size compared to wounds infected with Pf negative strains. Together, these data implicate Pf phage in the delayed wound healing associated with Pa infection through direct manipulation of mammalian target cells. We propose that Pf phage may have potential as a biomarker and therapeutic target in delayed wound healing.

microbiology↗