HMMSplicer

HMMSplicer identifies splice junctions from RNA-Seq short reads to map exon-exon boundaries, including novel and inexact splicing events, to support transcriptome and alternative splicing analysis.

Key Features:

• HMM-based prediction: Employs a Hidden Markov Model to predict splice junctions from RNA-Seq reads and reports improved specificity and higher junction recovery across datasets including Arabidopsis thaliana, Plasmodium falciparum, and Homo sapiens.
• Splice-junction mapping: Maps short reads that span exon-exon junctions back to a reference genome to enable discovery of previously unknown junctions.
• Canonical and non-canonical junction detection: Identifies both canonical and non-canonical splice junctions.
• Inexact splicing detection: Detects inexact splicing events and in Homo sapiens identifies 3.6% of 3' splice sites and 1.4% of 5' splice sites as inexact, typically differing by three bases.
• Versatility across genomes: Effective in compact genomes and for genes with low expression, alternative splicing isoforms, or non-canonical splice junctions.
• Scoring system: Assigns a score to each predicted junction to allow thresholding between sensitivity and specificity.

Scientific Applications:

• Alternative splicing analysis: Supports identification and characterization of alternative splicing isoforms from RNA-Seq data.
• Transcriptome annotation and novel junction discovery: Enables discovery of novel splice junctions and transcriptome annotation in organisms lacking comprehensive gene models.
• Detection of rare and inexact events: Facilitates detection and analysis of rare and inexact splicing events, including non-canonical splice sites.
• Comparative transcriptomics: Applicable to transcriptomic studies in Arabidopsis thaliana, Plasmodium falciparum, and Homo sapiens, among other organisms.

Methodology:

Analyzes short reads from RNA-Seq data using a Hidden Markov Model to predict splice junctions without relying on pre-existing gene models; implemented in Python.