iFLAS: positive-unlabeled learning facilitates full-length transcriptome-based identification and functional...

Area of Science:

• Plant molecular biology
• Genomics
• Bioinformatics

Background:

• Full-length transcriptome sequencing accelerates novel splicing isoform discovery.
• Existing alternative splicing (AS) tools are not optimized for plant data, posing challenges for plant isoform identification and functional analysis.
• Plant and animal AS patterns exhibit significant differences.

Purpose of the Study:

• To develop an optimized toolkit for plant full-length alternative splicing analysis.
• To accurately identify novel isoforms and investigate their functions in plants.
• To enable differential AS, poly(A) tail length, and allele-specific AS (ASAS) analyses in plants.

Main Methods:

• Development of integrated full-length alternative splicing analysis (iFLAS), a plant-optimized AS toolkit.
• Implementation of a semi-supervised machine learning method (positive-unlabeled or PU learning) for novel isoform identification.
• Application of iFLAS to maize (Zea mays) full-length transcriptome sequencing datasets for systematic analysis.

Main Results:

• iFLAS accurately identified novel plant isoforms and enabled functional investigation.
• Intron retention was found to introduce premature termination codons and regulate 3'UTR/poly(A) tail length, affecting isoform function.
• Distinct allele-specific AS patterns were observed in maize heterosis offspring, suggesting potential breeding applications.

Conclusions:

• iFLAS is a broadly applicable tool for plant full-length transcriptome-based AS research.
• The study provides insights into the functional roles of AS, including intron retention and ASAS, in plants.
• Findings highlight the potential of AS analysis in plant breeding and functional genomics.