Biology subjects

Bliss, A.

Publications and source records attributed to Bliss, A..

3 recordsLinked to original sources

bioRxiv2025

Resilient Antarctic soil bacteria consume trace gases across wide temperature ranges

Polar desert soils host diverse microbial communities despite limited nutrients and frequent temperature and light fluctuations. Adapting to these extremes, most bacteria possess high-affinity hydrogenases and carbon monoxide dehydrogenases, enabling them to use atmospheric trace gases such as hydrogen (H2) and carbon monoxide (CO) to generate energy and fix carbon. Despite the foundational importance of this process in polar desert ecosystems, little is known about the thermal sensitivity of trace gas oxidation or how this process will respond to climate warming. Here, we show through in situ and ex situ incubations that H2 consumption is an exceptionally thermally resilient process that can occur from -20 to +75{degrees}C, at rates comparable to temperate ecosystems (peaking at 8.56 nmol H2 h-1 g dry soil-1 at 25{degrees}C). Temperature ranges of CO (-20 to 42{degrees}C) and CH4 (-20 to 30{degrees}C) oxidation are also wider than expected, though the pattern of thermal sensitivity conforms with general theory. Metagenomic analyses support these data, revealing that atmospheric H2 and CO oxidisers are widespread, diverse, and abundant, and suggesting most Antarctic bacteria function below their temperature optima for these processes. Modelling of seasonal temperatures across ice-free Antarctica under current and future emissions scenarios indicates that H2 and CO oxidation can occur year-round, increasing by up to 35% or 44%, respectively, by 2100. Our results indicate constitutive aerotrophic activity contributing to Antarctic ecosystem functioning and biodiversity across spatial and temporal scales, with further studies required to understand how it interacts with photosynthesis in a changing climate.

microbiology↗

bioRxiv2025

Benchmarking large language models for cell-free RNA diagnostic biomarker discovery

Large-language models (LLMs) can parse vast amounts of data and generate executable code, positioning them as promising tools for the development of biomarkers and classifiers from high-throughput omics data. Here, we benchmarked six LLMs, OpenAIs o3 and GPT-4o, Anthropics Claude Opus 4 and Claude 3.7 Sonnet, and Googles Gemini 2.5 Pro and Gemini 2.0 Flash, for disease classification based on plasma cell-free RNA (cfRNA) profiles obtained by RNA sequencing. We analyzed data from cohorts of children with Kawasaki disease (KD) or multisystem inflammatory syndrome in children (MIS-C), adults with active tuberculosis (TB) or other non-TB respiratory conditions, and individuals with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) or sedentary lifestyle. We assessed two tasks: (i) gene-panel design, where each LLM mined public knowledge to nominate diagnostic genes for use in machine learning (ML), and (ii) end-to-end modeling, where LLMs built an ML workflow directly from raw RNA-seq counts. In the first task, the LLM-derived panels captured canonical immune pathways and outperformed randomly selected genes in all cohorts. They underperformed panels chosen by differential gene expression (DGE) analysis in the KD vs. MIS-C and ME/CFS cohorts but performed comparably or better for the TB cohort. In the second task, o3 produced classifiers for KD vs. MIS-C that performed just as well as conventional statistical methods without human intervention. Performance for TB and ME/CFS cohorts was slightly lower than the conventional approach. These findings delineate current capabilities and limitations of LLMs in diagnostics and open a path for their future use in biomarker discovery.

bioinformatics↗

bioRxiv2025

Minimal Correlation but Complementary Diagnostic Utility for Plasma Cell-free RNA and Proteins

Proteins and RNA circulate in plasma and can offer insights into human physiology. Yet, despite their clinical importance, direct comparisons between these analytes remain unexplored. Here, we measured and compared plasma cell-free RNA (cfRNA) and protein levels for 263 children diagnosed with inflammatory diseases by RNA-sequencing (n=155) and SomaScan proteomics (n=171). Remarkably, cfRNA and protein levels were largely uncorrelated across samples (feature-by-sample r=0.052; median feature-level r=0.009). Nonetheless, machine learning models based on either modality distinguished Kawasaki Disease (KD) from Multisystem Inflammatory Syndrome in Children (MIS-C) with similar high accuracy (median AUC > 0.93). Analysis of KD subtypes revealed distinct cfRNA and protein signatures, with one group showing molecular similarity to MIS-C. These findings underscore the complementary nature of cfRNA and protein profiling and highlight the utility of integrating multiple blood analytes to improve disease classification and deepen our understanding of complex inflammatory conditions.

genomics↗