Biology subjects

Quan, P.

Publications and source records attributed to Quan, P..

3 recordsLinked to original sources

bioRxiv2026

SpatialOmicsLab: an integrated research environment for AI co-scientists in spatial transcriptomics

Spatial profiling is moving from method development into a central platform for biological discovery and clinical translation, but analyzing such complex data has become the bottleneck of scientific discovery. Translating spatial transcriptomic measurements into biological insight requires reproducible workflows that integrate diverse analytical methods. Although language-model agents can plan and execute multistep analyses, fragmented data and software ecosystems hinder their use as reliable research assistants. In this study, we introduce SpatialOmicsLab, an integrated research environment for spatial transcriptomics comprising 113 curated computational tools, 63 reference datasets and 4,614 task specifications. We also introduce ST-Coscientist, an AI co-scientist that uses this environment to assemble, execute and refine spatial transcriptomic analyses. Within SpatialOmicsLab, ST-Coscientist outperformed expert-led workflows and other scientific agents across six benchmark settings spanning spatial clustering, cell-type deconvolution and spatially variable gene detection. It constructed alignment-aware workflows for three-dimensional MERFISH data and addressed biological questions by selecting and analysing relevant reference datasets. Procedural memory increased the success rate of integrating and validating additional analytical tools from 48% to 95%. Together, SpatialOmicsLab and ST-Coscientist establish an extensible ecosystem for reproducible spatial transcriptomic research, enabling AI co-scientists to address evolving biological questions with an expanding analytical repertoire.

bioinformatics↗

bioRxiv2026

Spatial Topology Reveals Biologically Distinct Recurrent Motifs in Colorectal Cancer

Most spatial transcriptomic analyses of solid tumors focus on individual cell states or single-sample spatial domains rather than on recurrent multicellular tissue architectures shared across patients, and typically depend on predefined cell-type annotations or compartment definitions. We developed STORM (Spatial Topology analysis of Recurrent Motifs), an unsupervised graph-attention variational autoencoder that learns recurrent spatial motifs directly from cell-level graph structure and molecular profiles without cell-type labels, manual annotation, or predefined compartments. We applied to 32 Xenium sections from 16 patients with paired early-onset (EOCRC) and average-onset (AOCRC) colorectal cancer, STORM identified 10 recurrent motifs that self-organized into tumor-parenchymal, stromal, and immune macro-compartments. Among these, the Desmoplastic Fibrotic Barrier (DFB) motif, a CAF- and ECM-rich boundary architecture, was associated with restricted CD8+ T-cell geodesic access to tumor cores independently of CD8+ abundance, as demonstrated by abundance-normalized neighborhood enrichment statistics and within-sample mixed-effects models. EOCRC selectively amplified this barrier-exclusion architecture, exhibiting tighter tumor parenchyma, denser DFB shells, and a DFB-specific ECM activation program that yielded an age-specific prognostic signature in TCGA-COAD. Translation of motif macro-classes to H&E images via a Vision Transformer classifier produced an image-derived DFB-barrier composite that predicted overall survival in advanced-stage TCGA colorectal cancer. STORM provides an annotation-free framework for discovering recurrent spatial motifs and identifies a fibroblast barrier architecture whose topological association with immune exclusion is independent of effector abundance, amplified in early-onset disease, and translatable to a deployable pathology-based prognostic biomarker.

bioinformatics↗

bioRxiv2019

Reconciling the potentially irreconcilable? Genotypic and phenotypic amoxicillin-clavulanate resistance in Escherichia coli

Resistance to amoxicillin-clavulanate, a widely used beta-lactam/beta-lactamase inhibitor combination antibiotic, is rising globally, yet susceptibility testing remains challenging. To test whether whole-genome sequencing (WGS) could provide a more reliable assessment of susceptibility than traditional methods, we predicted resistance from WGS for 976 E. coli bloodstream infection isolates from Oxfordshire, UK, comparing against phenotypes from the BD Phoenix (calibrated against EUCAST guidelines). 339/976 (35%) isolates were amoxicillin-clavulanate resistant. Predictions based solely on beta-lactamase presence/absence performed poorly (sensitivity 23% (78/339)) but improved when genetic features associated with penicillinase hyper-production (e.g. promoter mutations, copy number estimates) were considered (sensitivity 82% (277/339); p<0.0001). Most discrepancies occurred in isolates with peri-breakpoint MICs. We investigated two potential causes; the phenotypic reference and the binary resistant/susceptible classification. We performed reference standard, replicated phenotyping in a random stratified subsample of 261/976 (27%) isolates using agar dilution, following both EUCAST and CLSI guidelines, which use different clavulanate concentrations. As well as disagreeing with each other, neither agar dilution phenotype aligned perfectly with genetic features. A random-effects model investigating associations between genetic features and MICs showed that some genetic features had small, variable and additive effects, resulting in variable resistance classification. Using model fixed-effects to predict MICs for the non-agar dilution isolates, predicted MICs were in essential agreement ({+/-}1 doubling dilution) with observed (BD Phoenix) MICs for 691/715 (97%) isolates. This suggests amoxicillin-clavulanate resistance in E. coli is quantitative, rather than qualitative, explaining the poorly reproducible binary (resistant/susceptible) phenotypes and suboptimal concordance between different phenotypic methods and with WGS-based predictions.

microbiology↗