Identification of kidney cell types in scRNA-seq and snRNA-seq data using machine learning algorithms.

Area of Science:

• Genomics
• Bioinformatics
• Nephrology

Background:

• Single-cell RNA sequencing (scRNA-seq) and single-nucleus RNA sequencing (snRNA-seq) offer deep insights into kidney cell biology.
• Manual cell type annotation is labor-intensive and limits research scalability.

Purpose of the Study:

• To evaluate the efficacy of supervised machine learning algorithms for automated kidney cell type annotation.
• To compare the performance of five distinct machine learning models.

Main Methods:

• Analysis of 62,120 cells from 79 kidney biopsy samples across five sc/snRNA-seq datasets.
• Application of five supervised algorithms (SVM, Random Forest, MLP, KNN, XGBoost) for cell type annotation.
• Evaluation using F1 scores and rejection rates on integrated and harmonized datasets.

Main Results:

• All five machine learning algorithms achieved high accuracy, with a median F1 score of 0.94.
• Algorithms demonstrated robust performance across datasets and effectively rejected unannotated cell types.
• Slightly reduced accuracy was observed when models trained on scRNA-seq data were applied to snRNA-seq data.

Conclusions:

• Machine learning provides an accurate and scalable method for annotating kidney cell types in sc/snRNA-seq data.
• This automated approach can standardize annotation processes and accelerate kidney disease research.
• Further validation with larger sample sizes is warranted.