Knowledge-based reconstruction of mRNA transcripts with short sequencing reads for transcriptome research

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Transcriptome analysis often focuses on gene-level expression, overlooking the diversity generated by alternative splicing and gene transcripts.
• Discovering novel mRNA transcripts is crucial for understanding genome functionality but presents bioinformatic challenges due to short sequencing reads and incomplete transcript coverage.

Purpose of the Study:

• To develop and validate a computational approach for reconstructing novel mRNA transcripts from short sequencing reads.
• To improve the annotation of gene transcripts by integrating prior knowledge from genomic databases.

Main Methods:

• A computational method was developed to reconstruct mRNA transcripts using short sequencing reads and reference annotations from databases like RefSeq.
• Prior knowledge of known transcripts was used to define exon boundaries and infer unsequmented transcript regions.
• The approach was tested on a deep sequencing dataset from human muscle tissue.

Main Results:

• The method identified 2,973 novel junctions, 7,471 exons, and 7,571 previously unannotated transcripts in RefSeq.
• A significant portion (73%) of these novel transcripts were supported by other major gene annotation databases.
• Reconstructed transcripts were substantially longer and more complete than those generated by de novo methods.

Conclusions:

• Incorporating existing transcript annotations from genomic databases significantly aids the reconstruction of novel transcripts from short sequencing reads.
• This approach enhances transcriptome research by enabling more comprehensive genome-wide studies and improving downstream applications like microarray design and isoform expression analysis.