Finding the active genes in deep RNA-seq gene expression studies.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• RNA sequencing (RNA-seq) initially suggested a vast transcriptome, but low-abundance transcripts were often considered noise.
• Previous RNA-seq analyses used arbitrary expression thresholds (e.g., 0.3-1 FPKM) due to limitations in sequencing depth and analytical methods.
• Advances in RNA-seq technologies and bioinformatics have overcome many previous uncertainties.

Purpose of the Study:

• To re-evaluate the accuracy and efficiency of RNA sequencing experiments.
• To differentiate between active and noise-driven transcripts in the mammalian transcriptome.
• To establish a robust method for quantifying gene expression.

Main Methods:

• Utilized genomic data from large-scale studies, including the ENCODE project, for validation.
• Developed a novel normalization metric, zFPKM, to distinguish active from background gene expression.
• Analyzed chromatin state data to correlate gene expression levels with promoter activity.

Main Results:

• Demonstrated that the human transcriptome comprises active genes and noise-associated transcripts.
• Showed ultralow-expression genes are linked to repressed chromatin.
• The zFPKM metric effectively separates active genes (associated with active promoters) from noisy, ultralow-expression genes (with repressed promoters).

Conclusions:

• The zFPKM normalization accurately identifies biologically relevant genes.
• A read depth of 20-30 million mapped reads ensures high-confidence quantitation at the determined threshold.
• ENCODE chromatin state data can be used to validate RNA-seq analysis pipelines.