With the emergence of single-cell RNA sequencing (scRNA-seq) technology, scientists are able to examine gene expression at single-cell resolution. Analysis of scRNA-seq data has its own challenges, which stem from its high dimensionality. The method of machine learning comes with the potential of gene (feature) selection from the high-dimensional scRNA-seq data. Even though there exist multiple machine learning methods that appear to be suitable for feature selection, such as penalized regression, there is no rigorous comparison of their performances across data sets, where each poses its own challenges. Therefore, in this paper, we analyzed and compared multiple penalized regression methods for scRNA-seq data. Given the scRNA-seq data sets we analyzed, the results show that sparse group lasso (SGL) outperforms the other six methods (ridge, lasso, elastic net, drop lasso, group lasso, and big lasso) using the metrics area under the receiver operating curve (AUC) and computation time. Building on these findings, we proposed a new algorithm for feature selection using penalized regression methods. The proposed algorithm works by selecting a small subset of genes and applying SGL to select the differentially expressed genes in scRNA-seq data. By using hierarchical clustering to group genes, the proposed method bypasses the need for domain-specific knowledge for gene grouping information. In addition, the proposed algorithm provided consistently better AUC for the data sets used.
Summary: Genome-wide association studies (GWAS) have been widely used to reveal the associations between genetic variations and phenotypes in a population of individuals. However, they have been criticized for missing important genetic markers usually due to the fact that the data may not fit the statistical models well. In this study, we address the challenge of identifying significant single nucleotide polymorphisms (SNPs) in GWAS by harnessing the capabilities of two sophisticated regression models, BIGLASSO and AUTALASSO. They are both variants of the least absolute shrinkage and selection operator (LASSO). Our research contributes to the field of genomics through detailed comparative analysis of Arabidopsis thaliana, revealing how each method specializes in uncovering SNPs for different trait types. Our findings indicate that BIGLASSO shows stronger alignment with GWAS results, particularly excelling in the analysis of binary traits, even when these are derived from categorical phenotypes. AUTALASSO could be effective for quantitative traits and complement GWAS. We demonstrate that these LASSO-based methods can significantly enhance the identification of genetic markers, offering a potent complement to traditional GWAS approaches. Our findings not only bridge the gap between statistical and machine learning methodologies in genetic studies but also provide a practical framework for researchers seeking to validate reported SNPs or explore new genomic regions for trait association. This work stands as a pivotal step toward the integration of advanced computational techniques in genomics, paving the way for more precise and comprehensive genetic analyses. Availability and implementation: Key results from the paper are available at the https://github.com/DongdongHou006/LASSO-SNP. The program was implementated using Python and R, and was tested using the Digital Research Alliance of Canada.
This research paper explores the use of ChatGPT in database management. ChatGPT, an AI-powered chatbot, has limitations in performing tasks related to database management due to the lack of standardized vocabulary and grammar for representing database semantics. To address this limitation, the paper proposes a solution that involves developing a set of syntaxes that can represent database semantics in natural language. The syntax is used to convert database schemas into natural language formats, providing a new application of ChatGPT in database management. The proposed solution is demonstrated through a case study where ChatGPT is used to perform two tasks, semantic integration, and tables joining. Results demonstrate that the use of semantic database representations produces more precise outcomes and avoids common mistakes compared to cases with no semantic representation. The proposed method has the potential to speed up the database management process, reduce the level of understanding required for database domain knowledge, and enable automatic database operations without accessing the actual data, thus illuminating privacy protection concerns when using AI. This paper provides a promising new direction for research in the field of AI-based database management.
Accurate prioritization of T-cell receptor (TCR)-epitope interactions and identification of tumor-reactive T cells are important but difficult steps in immunotherapy-oriented bioinformatics workflows. Existing methods typically address these tasks separately and either model TCR-epitope pairs as independent observations or rely primarily on transcriptomic signatures. In this study, we present DUET (Dual Unified Evaluation of TCR-Epitopes and Tumor-reactive T cells), a graph-based computational workflow that unifies both applications within a single heterogeneous graph framework. The protocol represents TCRs, epitopes, and T cells as typed nodes connected by similarity and association edges, and combines pretrained sequence embeddings with edge-aware graph attention, Laplacian positional encoding, and bidirectional cross-domain attention. Applied to the IEDB and VDJdb benchmarks, DUET achieved AUROC/AUPR values of 0.937/0.922 and 0.992/0.990, respectively, outperforming five state-of-the-art algorithms under standard evaluation. On a single-cell RNA-seq tumor-reactivity benchmark, the workflow achieved an area under the receiver operating characteristic curve of 0.985 and an area under the precision-recall curve of 0.975, substantially exceeding transcriptomic signature-based baselines. Additional generalization analyses showed that DUET’s clearest graph-specific benefit occurred under epitope-disjoint TCR-epitope prediction. Ablation analysis showed that Laplacian positional encoding provided the largest performance gain, particularly in sparse graph settings. These results suggest that heterogeneous graph modeling can serve as a practical protocol for integrating receptor sequence, antigen context, and cellular phenotype in computational immunology.