bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.64898/2026.10.16.759703

Survival-associated cellular communication networks in human solid cancers

Lee, D.·Bharadwaj, V.·Liang, J.·Lill, G. R.·Ji, J. H.·Campbell, E.·Ho, K.·O'Sullivan, T.
View original source ↗Export metadata ↓

Abstract

Single-cell atlases have enabled the prediction of human cell-cell communication (CCC) networks in cancer, but whether predicted communication networks have clinical significance is not well understood. Here, we integrated a pan-solid cancer scRNA-seq atlas comprising 78 cell types from 318,426 quality-controlled cells with bulk RNA-seq and clinical outcome data from The Cancer Genome Atlas (TCGA) to develop Single-cell-based Kaplan-Meier Analysis Network (SKyNet). SKyNet prioritizes ligand-receptor interactions associated with overall survival across 11 human solid cancers, predicting unique interactions not inferred by previous CCC algorithms. In pancreatic adenocarcinoma, SKyNet-predicted endothelial-pericyte interactions were associated with poor outcomes that were validated by independent spatial transcriptomics and survival cohorts. In lung adenocarcinoma, lymphatic endothelial cell interactions consisting of collagen-integrin pairs were associated with poorer survival outcomes. In kidney renal clear cell carcinoma and kidney renal papillary cell carcinoma, fibroblast communication networks displayed disparate survival associations, with the DSC2-DSG2 axis associated with better survival outcomes. Thus, SKyNet provides a hypothesis-generating framework for cell type-specific communication pathways associated with patient outcomes.

Explore related subjects

systems biologyPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Lee, D., Bharadwaj, V., Liang, J., Lill, G. R., Ji, J. H., Campbell, E., Ho, K., O'Sullivan, T.. 2026-10-19. Survival-associated cellular communication networks in human solid cancers. https://doi.org/10.64898/2026.10.16.759703

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Scalable proteomics with multiDIA resolves proteoforms in neurodegenerative disease

Deep proteome characterization has traditionally required multidimensional chromatographic fractionation, long acquisition times and large sample amounts, thus limiting scalability. Here, we introduce multiDIA, which combines independent digestions with multiple proteases and data-independent acquisition (DIA) to increase sequence coverage without extensive chromatographic fractionation. By generating complementary and overlapping peptide representations of proteins, multiDIA identifies about 270,000 distinct peptides per sample, including 3,649 phosphosites. This is more than twice the coverage of unfractionated workflows and exceeds the coverage of multidimensional fractionation workflows that use 10-fold more material and time. We demonstrate the approach in Parkinson's disease (PD) brain samples, showing that multiDIA substantially expands sequence coverage while retaining the sensitivity, reproducibility, and scalability of DIA. This enabled the identification of disease-associated proteoform changes undetectable by single-protease DIA and the quantification of many disease-relevant post-translational modifications. MultiDIA thus enables deep, high-confidence proteome characterization from limited sample quantities without the throughput and sensitivity constraints of multidimensional fractionation.

systems biology↗
bioRxiv2026

Genome-scale metabolic models for a synthetic soil microbial community as a path for understanding community functioning

Microbial communities drive essential ecosystem processes, yet community-level behavior often deviates from predictions based on single-species studies. This mismatch reflects the context-dependent nature of nutrient dependencies, metabolic interactions, and resource availability in complex environments. Here, we present a mechanistic, species-resolved framework to dissect metabolism and nutrient exchange in a defined 21-member soil synthetic community spanning five bacterial phyla. For each isolate, we reconstructed and manually curated thermodynamically constrained genome-scale metabolic models, reconciling predictions with growth phenotypes across rich and minimal media. We mapped minimal nutrient requirements, cataloged environment-specific uptake and secretion profiles, and compared metabolic pathways to identify potential competitive and cooperative interactions. Using community-scale metabolomics, we assigned putative producers and consumers for detected metabolites and revealed medium-dependent shifts in niche overlap. This approach exposes metabolic strategies that govern community function and resilience, offering a predictive basis for understanding and manipulating microbiomes. Our findings highlight how species-level metabolic maps can clarify why microbial interventions succeed or fail and guide targeted manipulations in soil ecosystems.

systems biology↗
bioRxiv2026

Genetic Perturbation Prediction Models Fail Out-of-Distribution: A Rigorous Benchmark and Systematic Evaluation

Perturbation prediction models - which forecast transcriptional responses to gene inactivation - are increasingly proposed as in silico surrogates for genome-scale drug discovery screens. Yet, rigorous evaluation of whether these models genuinely generalize beyond their training distribution has been lacking. Existing benchmarks rely predominantly on random train-test splits, obscuring systematic failure modes that only emerge under realistic out-of-distribution (OOD) conditions. Here we introduce EL-PERTURB (Extrapolative Learning for PERTURBation Prediction), a benchmarking framework that stress-tests 20 perturbation prediction models spanning six architecture classes on both essential gene and genome-wide Perturb-seq screens, using a novel stratified OOD splitting strategy that partitions train and test sets into functionally distinct regions of transcriptional space. We find that model performance degrades substantially under OOD conditions, and that architectural complexity - including single-cell foundation model embeddings - confers no reliable advantage over well-tuned simpler baselines. EL-PERTURB establishes a rigorous benchmark for perturbation effect prediction across a gradient of generalization difficulty, providing a guide for building models that can genuinely serve as surrogates for wet-lab screening.

systems biology↗