Biology subjects

Smith, D. A.

Publications and source records attributed to Smith, D. A..

5 recordsLinked to original sources

bioRxiv2025

Exposing the Molecular Reaction Blind Spots of LLMs with PathwayQA

Proteins mediate a large portion of cellular activity, and understanding protein pathways can yield novel biological insights. Large language models (LLMs) have become increasingly adept at performing inference tasks across different fields of science and engineering. These models could facilitate the analysis of protein networks and help generate hypotheses about protein interactions in a scalable and accessible manner. However, the performance of LLMs in inferring protein-mediated biochemical reactions remains understudied. Here, we evaluate nine LLMs in reasoning over protein pathways included in the curated Reactome database. We find that all nine models struggle to infer products of a reaction when given reactants and enzymes. GPT-4o mini performed the best with a median recovery score of 0.6667, but no model surpassed the baseline strategy of parroting reactants back as predicted products. Most LLMs also performed poorly when inferring whether a protein pathway is associated with a human disease, with an average accuracy of 0.5980. DeepSeek 7B Chat performed the best with an accuracy of 0.9100. This study highlights an area where LLMs still struggle to make correct inferences and provides an opportunity for further work in developing biological LLMs. We also provide a novel question-answer dataset, PathwayQA, which is based on the Reactome database. PathwayQA can be used to benchmark and improve model performance on reasoning over protein-interaction networks. PathwayQA is available at https://github.com/Helix-Research-Lab/PathwayQA.

bioinformatics↗

bioRxiv2025

Isolation and characterization of TayeBlu, a novel bacteriophage of Azotobacter vinelandii

Soil microbial communities drive global biogeochemical cycles and alter crop yields through nitrogen fixation. As agents of genetic mobility, mortality, and nutrient release, viruses have been shown to influence microbial community structure and activity in numerous marine and aquatic systems. However, their impacts on terrestrial ecosystems are less well understood, in part because few model phage-host systems have been established for soils. To fill this gap, we sought to develop a new model system for viral infection of nitrogen-fixing bacteria derived from agricultural soil. Here, we report the isolation, characterization, and sequencing of the novel bacteriophage TayeBlu, which infects the globally distributed aerobic soil bacterium Azotobacter vinelandii, a facultative diazotroph. TayeBlu was isolated from the rhizosphere of tomato plants in a farm greenhouse. We find that the availability of nitrogen to host cells strongly influences TayeBlu infection physiology at the level of adsorption kinetics, time to lysis, and burst size. Taxonomic and comparative genome analysis reveal that TayeBlu belongs to an understudied family in class Caudoviricetes in which a small core of structural and assembly genes has persisted through adaptive diversification on different bacterial hosts. IMPORTANCEUnderstanding the forces regulating soil microbial activity is critical for building accurate ecosystem models that can inform land-management strategies aimed at mitigating climate risks and stabilizing the global food supply. For agricultural sustainability, it is particularly important to understand the dynamics of soil nitrogen-fixing bacteria like Azotobacter vinelandii, a well-studied and globally distributed species whose activity promotes plant growth and soil fertility. To support detailed investigations of the impact of viruses on diazotroph ecosystem outputs, we isolated and investigated a novel soil virus that infects A. vinelandii. This new phage-host system holds promise as a model experimental system for soil viral studies, illuminating a critical but poorly understood aspect of soil ecology.

microbiology↗

bioRxiv2025

Molecular mechanisms of plastic biodegradation by the fungus Clonostachys rosea

Microbial degradation provides an avenue for the remediation of select plastic polymers contributing to the urgent environmental problem of global plastic pollution. We demonstrate the degradation of polycaprolactone (PCL) by Clonostachys rosea and elucidate its underlying molecular mechanisms. We constructed the genome of this fungal strain and monitored changes in gene expression when exposed to PCL. Twelve genes linked to PCL degradation were found in the genome of C. rosea, and some of them were upregulated in the presence of the plastic, including genes coding for two cutinases. We heterologously expressed the enzymes coded by both genes and confirmed their activity against PCL polymers. We also demonstrate that one of the enzymes was active against polyethylene terephthalate (PET) polymers. Glucose inhibited the expression of both genes, completely halting the plastic biodegradation process, possibly serving as a preferred and readily metabolisable carbon source compared with PCL. We confirm the presence of key metabolic pathways linked to PCL degradation in C. rosea, including fatty acid degradation, providing further evidence of the mechanisms central to plastic biodegradation.

microbiology↗

bioRxiv2020

Viral genome wide association study identifies novel hepatitis C virus polymorphisms associated with sofosbuvir treatment failure

Chronic hepatitis C virus (HCV) infection is a major cause of chronic liver disease, cirrhosis and hepatocellular carcinoma worldwide. With the recent development of direct acting antivirals (DAA), treatment of chronically infected patients has become highly effective although a subset of patients do not respond to therapy. Sofosbuvir is a common component of current de novo or salvage combination therapies. We used pre-treatment whole genome sequencing of HCV from 507 patients infected with HCV subtype 3a and treated with sofosbuvir containing regimens to detect viral polymorphisms associated with response to treatment. We found that three common polymorphisms present in HCV NS2 and NS3 proteins (not direct targets of sofosbuvir) were associated with reduced treatment response. These polymorphisms were enriched in post-treatment HCV sequences of patients unresponsive to treatment; they were also associated with lower reductions in viral load in the first week of therapy. The finding of polymorphisms in NS2 and NS3 proteins associated with poor treatment outcomes emphasises the value of more systematic genome-wide analyses of HCV in uncovering indirect but clinically relevant mechanisms of antiviral resistance.

microbiology↗

bioRxiv2018

DNA barcodes for rapid, whole genome, single-molecule analyses

We report an approach for visualizing DNA sequence and using these DNA barcodes to search complex mixtures of genomic material for DNA molecules of interest. We demonstrate three applications of this methodology; by identifying specific molecules of interest from a dataset containing gigabasepairs of genome; by straightforward strain-typing of bacteria from such a dataset and, finally, by locating infecting virus molecules in a background of human genomic material. DNA barcodes enable quantitative understanding of complex genome mixtures, on a large scale. As a result of the dense fluorescent labelling of the DNA, individual barcodes of the order 40 kilobase pairs in length can be reliably identified. This means DNA can be prepared for imaging using standard handling and purification techniques. The recorded dataset provides stable physical and electronic records of the total genomic content of a sample that can be readily searched for a molecule or region of interest.

genomics↗