Biology subjects

Shaw, A. G.

Publications and source records attributed to Shaw, A. G..

7 recordsLinked to original sources

bioRxiv2026

BiomniBench: Process-level Evaluation of LLM Agents for Real-world Biomedical Research

LLM agents now perform real biomedical research, but evaluating them rigorously is hard. Outcome-only benchmarks fail in two ways. First, a correct final answer can come from memorization, reward hacking, or wrong reasoning that produces the right number by chance. Second, valid alternative analyses are marked wrong simply because they differ from the reference. We introduce BiomniBench, a process-level evaluation framework that scores the full agent trajectory against expert-designed, task-specific rubrics. Our first release, BiomniBench-DA, contains 100 data-analysis tasks across 17 task types, 5 disease areas, and a general-biology category, each based on a paper from journals such as Nature, Cell, and Science and co-developed with an original author or a domain expert. Benchmarking frontier and open-weight models across four agent harnesses reveals three findings. Frontier and open-weight bases cluster within a few points of each other, with substantial headroom for all models. The agent harness shifts scores by more than the gap between successive model generations. Agents reliably ground claims in real sources yet consistently fall short on method selection, biological interpretation, and scientific reasoning. BiomniBench is the first process-level benchmark for LLM agents in biomedical research, providing the dimension-level diagnostics that outcome scoring cannot. Datasethuggingface.co/datasets/phylobio/BiomniBench-DA

bioinformatics↗

bioRxiv2025

Multiscale Spatial Mapping of Microbial Communities for Biotherapeutic Development

Live biotherapeutic products (LBPs) are emerging as powerful tools to modulate the microbiome using well-defined microbial communities. Yet, designing, manufacturing, and delivering LBPs remains challenging, in part due to a lack of technologies capable of analyzing LBPs as complete, spatially organized consortia. Conventional sequencing-based methods lack sensitivity and specificity and do not provide critical spatial information. To address this, we present high-phylogenetic-resolution spatial mapping platform (HiPR-Map), a state-of-the-art spectral imaging technology that enables precise enumeration and spatial localization of microbial cells at species-level within complex communities. Through these advantages, HiPR-Map provides unique insights for LBP discovery and development. Applying HiPR-Map to an LBP designed to complement immune checkpoint therapy, we profiled over 1.8 million microbial cells engrafted in the murine gut. Our analysis revealed distinctive microbial spatial organization, underscoring the power of imaging-based microbiome profiling to optimize LBP design and characterization. This work highlights the transformative potential of spatial microbiome analysis for next-generation LBP development.

microbiology↗

bioRxiv2025

Staphylococcus haemolyticus is a reservoir of antibiotic resistance genes in the preterm infant gut

Among coagulase-negative staphylococci, Staphylococcus haemolyticus is a primary cause of bloodstream infections in preterm infants, with gut colonisation being recognised as a risk factor for subsequent infection. Through a re-analysis of a 16S rRNA gene sequencing dataset (n=497 preterm infants), we found that S. haemolyticus was abundant and prevalent in the gut in the first month of life. To better understand the diversity of S. haemolyticus among preterm infants, we generated genome sequences of S. haemolyticus strains (n=140), which were isolated from 44 stool samples of 22 preterm infants from four different hospitals in the United Kingdom. Core genome phylogenetic analyses, incorporating 126 publicly available S. haemolyticus genome sequences, showed that 85/140 (60.1%) of the isolates, from three different hospitals, formed a clonal group with 79/85 (92.9%) strains being assigned to Multi-Locus Sequence Type (ST) 49. Antibiotic resistance genes were highly prevalent in the genome sequences. Using logistic regression, we found a strong association between the presence of the gene mecA and phenotypic resistance to oxacillin (odds ratio [OR]: 158.00, p<0.0001), and the aacA-aphD gene and phenotypic resistance to gentamicin aacA-aphD (OR: 162.00, p<0.001). None of the strains from the preterm infant cohort had a complete Staphylococcal Cassette Chromosome mec (SCCmec) element. The aacA-aphD gene was associated with the transposon Tn4001. Using hybrid genome assemblies, we found it to be present on the chromosome (54.5% of strains) or on diverse plasmids (27.3%). Four strains (18.2%) had Tn4001 copies on both plasmid and chromosome. Our data suggest the existence of a distinct sub-population of S. haemolyticus that has adapted to colonise the gut of preterm infants. Prevalent resistance to antibiotics is of clinical concern and the diversity of genetic contexts of mecA and Tn4001 suggests widespread horizontal gene transfer and recombination in this species.

microbiology↗

bioRxiv2024

Impact of early life antibiotic and probiotic treatment on gut microbiome and resistome of very-low-birth-weight preterm infants

Preterm infants (<37 weeks gestation) are often administered broad-spectrum antibiotics in hospitals due to their vulnerability to severe morbidity, including necrotising enterocolitis and sepsis. However, antibiotics can disrupt the development of early-life microbiota, potentially impairing gut immunity and colonisation resistance. Evidence shows that probiotics (e.g., certain Bifidobacterium strains) may help restore healthy gut microbiota. In this study, we examined the effects of probiotics and antibiotics on the preterm gut microbiome and resistome in two unique cohorts of 34 very-low-birth-weight, human-milk- fed preterm infants (moderate to very preterm), with one cohort receiving probiotics. Within each group, some infants were treated with antibiotics (benzylpenicillin and/or gentamicin) while others served as non-antibiotic treated controls. We performed shotgun metagenomic sequencing on 93 longitudinal faecal samples from 34 infants, generated >300 metagenome- assembled genomes, and obtained [~]90 isolate genomes through targeted culturomics, enabling analysis of the microbiome/resistome at species and strain levels. Additionally, we investigated in vitro horizontal gene transfer (HGT) capacity of preterm infant-derived multidrug-resistant (MDR) pathogen Enterococcus via neonatal gut models. Overall, probiotic supplementation significantly reduced antibiotic resistance gene prevalence, MDR pathogen load, and helped restore a typical early-life microbiota. However, the persistence of MDR pathogens like Enterococcus, with high HGT potential, highlights the need for ongoing surveillance in neonatal care. Our findings underscore the complex interactions between antibiotics, probiotics, and HGT in shaping the neonatal microbiome and support further research into probiotics for antimicrobial stewardship in preterm populations.

microbiology↗

bioRxiv2023

Automated detection and classification of polioviruses from nanopore sequencing reads using piranha

Widespread surveillance, rapid detection and appropriate intervention will be critical for successful eradication of poliovirus. With deployable next-generation sequencing (NGS) approaches, such as Oxford Nanopore Technologies MinION, the time from sample to result can be significantly reduced compared to cell culture and Sanger sequencing. We developed piranha (poliovirus investigation resource automating nanopore haplotype analysis) to aid routine poliovirus testing of both stool and environmental samples and alleviate the bioinformatic bottleneck that often exists for labs adopting novel NGS approaches. Piranha can be used for efficient intratypic differentiation of poliovirus serotypes, for classification of Sabin-like polioviruses and for detection of wild-type and vaccine derived polioviruses. It produces interactive, distributable html reports, as well as summary csv files and consensus poliovirus FASTA sequences. The reports describe each nanopore sequencing run with interpretable plots enabling researchers to easily detect the presence of poliovirus in samples and quickly disseminate their results.

bioinformatics↗

bioRxiv2022

Development of the gut microbiota during early life in premature and term infants

BackgroundThe gastrointestinal (GI) microbiota has been linked to health consequences throughout life, from early life illnesses (e.g. sepsis and necrotising enterocolitis) to lifelong chronic conditions such as obesity and inflammatory bowel disease. It has also been observed that events in early life can lead to shifts in the microbiota, with some of these changes having been documented to persist into adulthood. A particularly extreme example of a divergent early GI microbiota occurs in premature neonates, who display a very different GI community to term infants. Certain characteristic patterns have been associated with negative health outcomes during the neonatal period, and these patterns may prove to have continual damaging effects if not resolved. ResultsIn this study we contrast a set of premature infants with a paired set of term infants (n=37 pairs) at six weeks of life and at two years. In the samples taken at six weeks we find microbial communities differing in both diversity and specific bacterial groups between the two infant cohorts. We identify clinical factors associated with over-abundance of potentially pathogenic organisms (e.g. Enterobacteriaceae) and reduced abundances of some beneficial organisms (e.g. Bifidobacterium). We contrast these findings with samples taken at two years of age, which show that despite a very different initial microbiota, the two infant groups converge to a similar, more adult-like state. We identify clinical factors, including both prematurity and delivery method, that remain associated with components of the microbiota. Both clinical factors and microbial characteristics are compared to the occurrence of childhood wheeze and eczema, revealing associations between components of the GI microbiota and the development of these allergic conditions. ConclusionsThe faecal microbiota differs greatly between infants born at term and those born prematurely during early life, yet it converges over time. Despite this, early clinical factors remain significantly associated with the abundance of some bacterial groups at two years of age. Given the associations made between health conditions and the microbiota, factors that alter the makeup of the gut microbiota, and potentially its trajectory through life, could have important lifelong consequences.

genomics↗

bioRxiv2020

Rapid and sensitive direct detection and identification of poliovirus from stool and environmental surveillance samples using nanopore sequencing

Global poliovirus surveillance involves virus isolation from stool and environmental samples, intratypic differential (ITD) by PCR and sequencing of the VP1 region to distinguish vaccine (Sabin), vaccine-derived and wild-type polioviruses and ensure an appropriate response. This cell-culture algorithm takes 2-3 weeks on average between sample receipt and sequencing. Direct detection of viral RNA using PCR allows faster detection but has traditionally faced challenges related to poor sensitivity and difficulties in sequencing common samples containing poliovirus and enterovirus mixtures. We present a nested PCR and nanopore sequencing protocol that allows rapid (<3 days) and sensitive direct detection and sequencing of polioviruses in stool and environmental samples. We developed barcoded primers and a real-time analysis platform that generate accurate VP1 consensus sequences from multiplexed samples. The sensitivity and specificity compared with cell-culture were 90.9% (95% Confidence Interval: 75.7-98.1%) and 99.2% (95.5-100.0%) for wild-type 1 poliovirus, 92.5% (79.6-98.4%) and 98.7% (95.4-99.8%) for vaccine and vaccine-derived serotype 2 poliovirus, and 88.3% (81.2-93.5%) and 93.2% (88.6-96.3%) for Sabin 1 and 3 poliovirus alone or in mixtures when tested on 155 stool samples in Pakistan. Variant analysis of sequencing reads also allowed identification of polioviruses and enteroviruses in artificial mixtures and was able to distinguish complex mixtures of polioviruses in environmental samples. The median identity of consensus nanopore sequences with Sanger or Illumina sequences from the same samples was >99.9%. This novel method shows promise as a faster and safer alternative to cell-culture for the detection and real-time sequencing of polioviruses in stool and environmental samples.

microbiology↗