scFSNN: a feature selection method based on neural network for single-cell RNA-seq data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) provides high-resolution gene expression data.
• scRNA-seq data presents unique challenges including over-dispersion, zero-inflation, and high dimensionality.
• Existing feature selection methods struggle with the complexity of scRNA-seq data.

Purpose of the Study:

• To develop a robust feature selection method tailored for scRNA-seq data.
• To address the challenges of high dimensionality and complex data characteristics in scRNA-seq.
• To improve classification performance on scRNA-seq datasets.

Main Methods:

• A novel feature selection method based on neural networks, termed scFSNN, was developed.
• scFSNN is an embedded method that performs feature selection during model training.
• The method incorporates automatic feature selection, false discovery rate control, and adaptive feature elimination.

Main Results:

• scFSNN demonstrated superior feature selection capabilities compared to existing methods.
• The method achieved excellent predictive performance in classification tasks.
• Extensive simulations and real-world data analyses validated scFSNN's effectiveness.

Conclusions:

• scFSNN offers an effective solution for feature selection in scRNA-seq data analysis.
• The method enhances the accuracy and reliability of cell classification from scRNA-seq data.
• scFSNN provides a valuable tool for researchers working with complex single-cell genomics datasets.