Integrating machine learning and single-cell sequencing to reveal the role of kinase-related genes in subtype classification and prognostic significance of lung squamous cell carcinoma

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Lung squamous cell carcinoma (LUSC) represents a prevalent form of lung cancer that poses a significant threat to human health, characterized by a steadily rising global incidence and mortality rate, along with a dismal prognosis. Kinases, a category of enzymes responsible for catalyzing the phosphorylation of various substrates, are crucial in mediating cell signaling, regulating metabolic processes, controlling the cell cycle, and facilitating tumor progression. This research integrates machine learning techniques with single-cell sequencing to investigate the involvement of kinase-related genes (KRGs) in the classification of LUSC subtypes and their prognostic implications, thereby offering innovative perspectives for personalized treatment approaches in LUSC. Utilizing the TCGA-LUSC dataset, we identified a total of 4825 differentially expressed genes (DEGs). Through Cox regression analysis, we pinpointed 13 KRGs that exhibit potential prognostic relevance. Based on these KRGs, we categorized LUSC patients into two subtypes, C1 and C2. Concurrently, we leveraged the DEGs distinguishing the C1 and C2 groups to further classify patients into two additional subtypes (CA and CB). Notably, patients categorized within the C1 and CB groups exhibited a poorer prognosis compared to those in the C2 and CA groups. Furthermore, significant disparities were observed in functional enrichment and the immune microenvironment across these subtypes, with individuals in the C1 and CB groups demonstrating elevated levels of immune checkpoint expression relative to the C2 and CA cohorts. Additionally, through LASSO and Cox regression analyses, we identified four pivotal genes: ROS1, ACVRL1, LATS2, and CHEK2. Prognostic models based on these genes were developed and subsequently validated using two external datasets sourced from the GEO database. We also conducted a comprehensive analysis of the correlation between the riskscores from the prognostic models and immune cell infiltration, immune checkpoints, and responses to immunotherapy. Importantly, the single-cell sequencing analysis yielded valuable insights into the microenvironmental heterogeneity present in LUSC. In conclusion, we investigated the variances in chemotherapeutic drug sensitivities between high-risk and low-risk patient groups, aiming to offer promising avenues for the selection of chemotherapeutic regimens tailored for LUSC.

Article activity feed