Sicklick, J. K. | bioRxiv Science

bioRxiv2025

Single Nuclei-Derived Molecular Subtypes of Gastrointestinal Stromal Tumors Correlate with Clinicopathologic Features and Predict Clinical Outcomes

Historically, gastrointestinal stromal tumor (GIST) has been subtyped by oncogenic driver mutations. However, tumors with the same mutational profile can have variable biology. To further explore the impact of molecular diversity on GIST biology, we performed single nucleus RNA sequencing on 16 primary GIST and utilized an integrated single cell atlas of the normal GI tract composed from multiple publicly available datasets to identify six distinct GIST cell states. We then statistically estimated the relative abundances of these profiles in bulk transcriptomic data. These were used to define six common GIST molecular subtypes based upon one or two predominant tumor cell states. We found that these molecular subtypes correlate with tumor locations, mutational profiles, and patient outcomes, and validated these subtypes in an independent international cohort. These molecular subtypes have the potential to be used for clinical prognostication for patients with GIST, identifying new therapeutic targets, and studying the cell of transformation of GIST.

cancer biology↗

bioRxiv2023

NMFClustering: Accessible NMF-based clustering utilizing GPU acceleration

SummaryNon-negative Matrix Factorization (NMF) is an algorithm that can reduce high dimensional datasets of tens of thousands of genes to a handful of metagenes which are biologically easier to interpret. Application of NMF on gene expression data has been limited by its computationally intensive nature, which hinders its use on large datasets such as single-cell RNA sequencing (scRNA-seq) count matrices. We have implemented NMF based clustering to run on high performance GPU compute nodes using CuPy, a GPU backed python library, and the Message Passing Interface (MPI). This reduces the computation time by up to three orders of magnitude and makes the NMF Clustering analysis of large RNA-Seq and scRNA-seq datasets practical. We have made the method freely available through the GenePattern gateway, which provides free public access to hundreds of tools for the analysis and visualization of multiple omic data types. Its web-based interface gives easy access to these tools and allows the creation of multi-step analysis pipelines on high performance computing (HPC) clusters that enable reproducible in silico research for non-programmers. Availability and ImplementationNMFClustering is freely available on the public GenePattern server at https://genepattern.ucsd.edu. Code for the NMFClustering is available under a BSD style license on github at https://github.com/genepattern/nmf-gpu. ContactTed Liefeld, jliefeld@cloud.ucsd.edu Supplementary InformationSupplementary data are available at Bioinformatics online and at https://datasets.genepattern.org/?prefix=data/test_data/NMFClustering/.

bioinformatics↗