Vral, A. | bioRxiv Science

bioRxiv2026

Predicting subclonal TP53 mutations from tumor spatial transcriptomics data using a graph convolutional neural network

Spatial transcriptomics (ST) has revolutionized our understanding of tumor biology but inherently lacks information on the upstream somatic driver mutations. We developed a spatially-aware graph convolutional neural network (MuT-GCNN) that infers TP53 clones directly from ST data. MuT-GCNN was trained on virtual ST slides with clones simulated from a large collection of existing RNA and matched DNA sequencing data. The model is highly performant with precision and recall values exceeding 95% in most analysed cancer types. It is sensitive for single hit mutations and is primarily informed by the expression of p53 signalling genes in cancer cells. After demonstrating the potential of the model on publicly available squamous cell carcinoma (SCC) data, a direct validation was performed using ST and matched DNA sequencing from serial slices obtained from 4 cutaneous SCC samples. With the increasing availability of ST data and upcoming ST atlases, MuT-GCNN can unveil the location of (sub)clonal alterations in TP53, the most frequently mutated gene in human cancer.

cancer biology↗

bioRxiv2022

A clinically annotated post-mortem approach to study multi-organ somatic mutational clonality in histologically healthy tissues.

Recent research on histologically healthy human tissues identified omnipresent mutational microclones, driven by somatic mutations known to be responsible for carcinogenesis (e.g., in TP53 or NOTCH1). These new insights are fundamentally changing current tumour evolution models, with broad oncological implications. Most studies are based on surgical remnant tissues, which are not available for many organs and rarely in a pan-organ setting (multiple organs from the same individual). Here, we describe an approach based on clinically annotated post-mortem tissues, derived from whole-body donors that are routinely used for educational purposes at human anatomy units. We validated this post-mortem approach using UV-exposed and unexposed epidermal skin tissues and confirm the presence of positively selected NOTCH1/2-, TP53- and FAT1-driven clones. No selection signals were detected in a set of immune genes or housekeeping genes. Additionally, we provide the first evidence for smoking-induced clonal changes in oral epithelia, likely underlying the origin of head and neck carcinogenesis. In conclusion, the whole-body donor-based approach provides a nearly unlimited healthy tissue resource to study mutational clonality and gain fundamental mutagenic insights in the presumed earliest stages of tumour evolution.

genomics↗