همایش بین المللی هوش مصنوعی

صفحه اصلی / دومین همایش بین المللی هوش مصنوعی

Enhancing Single-Cell Transcriptomic Quality in Gastric Cancer: An Intersective Approach Combining Adaptive and Manual Quality Control

نویسندگان :

Fatemeh Khalaji-Pirbalouti¹ Modjtaba Emadi-Baygi²

1- دانشگاه شهرکرد 2- دانشگاه شهرکرد

کلمات کلیدی :

scRNA-seq،Quality control،Gastric cancer،Bioinformatics،Expression matrix

چکیده :

Single-cell RNA sequencing (scRNA-seq), a powerful genomic technology, has become crucial for analyzing cellular heterogeneity in complex diseases like gastric cancer (GC). The presence of technical artifacts, including ambient RNA and stressed or dying cells, necessitate stringent quality control (QC). We present a robust QC framework, implemented in the Python environment using Scanpy package, and applied to publicly available GC scRNA-seq dataset from the Gene Expression Omnibus (GEO) database. Our approach intersects adaptive QC, based on the median absolute deviation (MAD) for objective, outlier-based filtering, with manual QC, where thresholds are visually determined via histogram and violin plot inspection to remove low-quality cells. Specifically, we targeted three critical metrics: the number of detected genes, total unique molecular identifier counts (UMI), and the fraction of counts from mitochondrial genes per cell. Here, we profiled the transcriptomes of 134,367 cells from 28 patients with gastric cancer. By intersecting the set of cells removed by both methods, we derive a refined, high-quality cell population of 110,614 cells. This intersective methodology provides a flexible yet robust approach to mitigate the impact of technical noise, resulting in a cleaner count matrix optimized for downstream analysis of the GC tumor microenvironment. In conclusion, our intersective QC framework delivers a reproducible, high-quality gastric cancer scRNA-seq dataset by comprehensively mitigating technical artifacts. This dual-filtering strategy not only enhances the fidelity of cellular heterogeneity analyses but also establishes a robust foundation for uncovering molecular mechanisms within complex tumor microenvironments.