A heuristic model for computational prediction of human branch point sequence

Area of Science:

• Molecular Biology
• Bioinformatics

Background:

• Pre-messenger RNA (pre-mRNA) splicing removes introns and joins exons to form mature mRNA, a process critical for gene expression.
• The spliceosome, a complex molecular machine, catalyzes splicing. Splicing factor 1 (SF1) initially recognizes the branch point sequence (BPS), which is later bound by U2 snRNP.
• Predicting mammalian BPSs is challenging due to degenerate motifs and limited verified sequences.

Purpose of the Study:

• To develop an efficient heuristic model for predicting human branch point sequences (BPSs).
• To improve the accuracy of in silico BPS prediction by refining the search strategy and employing a novel scoring scheme.

Main Methods:

• Developed a heuristic model incorporating a novel scoring scheme to quantify the splicing strength of putative BPSs.
• Restricted candidate BPS identification to a defined region within introns to minimize interference from other splicing elements.
• Utilized two types of relative frequencies for human BPS prediction.

Main Results:

• The developed model demonstrated improved prediction accuracy for human BPSs.
• The heuristic model outperformed existing BPS prediction methods when tested on experimentally verified human introns.
• Genome-wide prediction revealed characteristics consistent with experimentally verified BPSs.

Conclusions:

• Binding energy is proposed as a contributing factor in the molecular recognition during human pre-mRNA splicing.
• The study provides a validated method for BPS prediction and a freely available webserver for its application.
• Predicted BPS characteristics and positions align with established findings in human pre-mRNA splicing research.