How Many Genes Are Expressed in a Transcriptome? Estimation and Results for RNA-Seq

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• RNA sequencing (RNA-seq) quantifies gene expression but can miss genes with low expression levels.
• Accurate estimation of the total number of expressed genes is crucial for understanding tissue-specific expression and transcriptome dynamics.
• Existing methods struggle to reliably estimate undetected genes and determine optimal sequencing depth.

Purpose of the Study:

• To develop a non-parametric estimator for the number of undetected genes in RNA-seq experiments.
• To provide a method for calculating the necessary sequencing depth to detect a target proportion of these undetected genes.
• To improve the accuracy and completeness of gene expression studies.

Main Methods:

• Developed a non-parametric statistical estimator inspired by ecological species estimation methods.
• Applied the estimator to 32 public RNA-seq experiments, analyzing 311 individual libraries.
• Compared the performance of the new estimator against previously published methods.

Main Results:

• The new estimators demonstrated reduced bias and smaller standard errors compared to existing methods.
• In most analyzed experiments, over a thousand genes remained undetected.
• On average, approximately 6% of expressed genes per accession were undetected, rising to 10% for individual libraries.

Conclusions:

• The developed method accurately estimates the number of undetected genes in RNA-seq data.
• It provides a reliable way to calculate the sequencing depth needed for more comprehensive gene detection.
• This approach enhances the accuracy of transcriptome-wide expression studies and is applicable to metagenomics.