Related Experiment Video
Updated: Jan 5, 2026
06:38
Isolation of Adult Spinal Cord Nuclei for Massively Parallel Single-nucleus RNA Sequencing
Published on: October 12, 2018
19.5K
SINC: a scale-invariant deep-neural-network classifier for bulk and single-cell RNA-seq data.
1Department of Applied and Computational Mathematics and Statistics, University of Notre Dame, Notre Dame, IN 46556, USA.
Bioinformatics (Oxford, England)
|October 25, 2019
Summary
This study introduces a scale-invariant deep neural network classifier (SINC) for RNA-seq data analysis. SINC bypasses sequencing depth normalization, improving classification accuracy and reliability, especially for single-cell RNA sequencing.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Accurate sequencing depth estimation is crucial for RNA-seq data analysis, but challenging for single-cell RNA sequencing (scRNA-seq).
- Inaccurate sequencing depth can compromise the validity of downstream analyses, including sample classification.
- There is a need for analysis methods that eliminate the reliance on sequencing depth, directly utilizing original count data.
Purpose of the Study:
- To develop a novel analysis method that is invariant to sequencing depth for RNA-seq data.
- To introduce a deep neural network-based classifier, named scale-invariant deep neural-network classifier (SINC), for sample classification tasks.
- To evaluate the performance of SINC against existing classifiers on diverse RNA-seq datasets.
Main Methods:
- Developed a scale-invariant (SI) analysis framework where results are independent of sequencing depth estimates.
- Designed a deep neural network architecture for the scale-invariant deep neural-network classifier (SINC).
- Applied SINC to nine bulk and single-cell RNA-seq datasets for sample classification.
Main Results:
- SINC achieved classification accuracy that was better than or competitive with eight other classifiers across nine datasets.
- The scale-invariant deep neural-network classifier (SINC) demonstrated superior ease of use and reliability, particularly on datasets with difficult-to-determine sequencing depths.
- The method effectively analyzes original count data without the need for sequencing depth normalization.
Conclusions:
- The scale-invariant deep neural-network classifier (SINC) offers a robust and accurate alternative for RNA-seq data classification.
- SINC mitigates issues associated with sequencing depth estimation, enhancing the reliability of scRNA-seq data analysis.
- The developed method provides a valuable tool for researchers in genomics and bioinformatics, particularly for comparative studies of normal versus cancerous samples.
