Moussa, M. R. | Search bioRxiv

bioRxiv2026

Predicting Gene Mutations in Colon Cancer Using Long-Term Temporal Dependency Learning on a Directed Co-Occurrence Asymmetry Graph

MotivationColorectal tumorigenesis follows stepwise mutational progression patterns, but inferring dependencies from mutation data remains challenging. We propose a directed co-occurrence asymmetry graph to infer graph-derived mutational paths from 2,344 colon adenocarcinoma samples covering 23,858 mutated genes. ResultsWe introduce a gene-conditioned shared Long Short-Term Memory (LSTM) model with attention to predict mutation status along inferred paths. The attention mechanism learns to weight informative predecessor mutations directly, while gene embeddings adapt the shared model to each target gene. We compare this model with standard per-gene LSTM and dilated Convolutional Neural Network (CNN) architectures. Graph-derived paths substantially improved precision and recall over the reproduced frequency-ordering baseline, with weighted paths performing best overall. The attention-based shared LSTM achieved competitive area under the ROC curve (AUC) and high recall, indicating that attention over predecessor mutations provides a useful representation for target-gene prediction. Code and datagithub.com/moussa-lab/MutationPrediction

bioinformatics↗

bioRxiv2025

A Benchmarking Study of Random Projections and Principal Components for Dimensionality Reduction Strategies in Single Cell Analysis

Principal Component Analysis (PCA) has long been a cornerstone in dimensionality reduction for high-dimensional data, including single-cell RNA sequencing (scRNA-seq). However, PCAs performance typically degrades with increasing data size, can be sensitive to outliers, and assumes linearity. Recently, Random Projection (RP) methods have emerged as promising alternatives, addressing some of these limitations. This study systematically and comprehensively evaluates PCA and RP approaches, including Singular Value Decomposition (SVD) and randomized SVD, alongside Sparse and Gaussian Random Projection algorithms, with a focus on computational efficiency and downstream analysis effectiveness. We benchmark performance using multiple scRNA-seq datasets including labeled and unlabeled publicly available datasets. We apply Hierarchical Clustering and Spherical K-Means clustering algorithms to assess downstream clustering quality. For labeled datasets, clustering accuracy is measured using the Hungarian algorithm and Mutual Information. For unlabeled datasets, the Dunn Index and Gap Statistic capture cluster separation. Across both dataset types, the Within-Cluster Sum of Squares (WCSS) metric is used to assess variability. Additionally, locality preservation is examined, with RP outperforming PCA in several of the evaluated metrics. Our results demonstrate that RP not only surpasses PCA in computational speed but also rivals and, in some cases, exceeds PCA in preserving data variability and clustering quality. By providing a thorough benchmarking of PCA and RP methods, this work offers valuable insights into selecting optimal dimensionality reduction techniques, balancing computational performance, scalability, and the quality of downstream analyses.

bioinformatics↗