Known sequence features explain half of all human gene ends.

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• Cleavage and polyadenylation (CPA) sites are crucial for defining eukaryotic gene termination.
• Key sequence elements, including UGUA, polyadenylation signal (PAS), U-rich, CA/UA, and downstream GU-rich elements (DSEs), are known to be involved in CPA.
• The sufficiency of these core elements in delineating functional CPA sites and the role of other factors remain unclear.

Purpose of the Study:

• To investigate whether the five primary sequence features are sufficient to delineate cleavage and polyadenylation (CPA) sites.
• To assess the contribution of individual sequence features to CPA site determination using discriminative models.
• To evaluate the impact of additional sequences and RNA-binding proteins (RBPs) on distinguishing functional from cryptic CPA sites.

Main Methods:

• Utilized standard discriminative models to analyze the contributions of individual sequence features to CPA.
• Developed models incorporating the five primary CPA sequence features.
• Assessed the performance boost from U1-hybridizing sequences and known RBP RNA binding motifs.

Main Results:

• Models using only the five primary CPA sequence features correctly identified constitutive CPA sites at the ends of coding genes for 59% of human genes.
• U1-hybridizing sequences offered a minor improvement in prediction accuracy.
• Incorporating all known RBP RNA binding motifs increased the prediction accuracy to only 61%, indicating a minimal role for these factors in distinguishing real from cryptic sites.

Conclusions:

• The five established sequence features are highly effective in predicting human gene ends, accounting for the majority of constitutive cleavage and polyadenylation (CPA) sites.
• Additional factors, including U1-hybridizing sequences and known RNA-binding protein motifs, contribute minimally to the accurate delineation of functional CPA sites.
• The core CPA machinery's sequence elements are largely sufficient for distinguishing functional gene ends from cryptic sites.