PsiCLASS

PsiCLASS assembles transcriptomes from RNA-seq reads using a reference-based approach to generate high-fidelity transcript models for gene expression quantification and downstream functional analyses.

Key Features:

• Simultaneous Multi-Sample Analysis: Analyzes single or multiple RNA-seq samples concurrently and applies mixture statistical models to improve exonic feature selection across datasets.
• Advanced Algorithmic Framework: Integrates splice graph-based dynamic programming algorithms with a weighted voting scheme for transcript selection.
• Superior Sensitivity-Precision Tradeoff: Reports substantially improved sensitivity-precision balance, with precision up to 2–3 fold higher than methods such as StringTie and the Scallop plus TACO combination.
• Efficiency and Scalability: Scales to large cohorts and assembles extensive datasets rapidly, demonstrated by assembling 667 GEUVADIS samples within 9 hours.
• Robust Accuracy Across Sample Sizes: Maintains robust assembly accuracy across varying and large numbers of samples.

Scientific Applications:

• Gene expression and functional analysis: Provides transcript models for quantification and downstream functional studies of gene expression.
• Comparative transcriptomics: Enables comparative analyses across multiple samples or conditions by joint assembly of RNA-seq datasets.
• Population-scale transcriptome assembly: Supports population genetics and large-cohort studies by assembling transcripts across many samples.
• High-throughput transcriptome reconstruction: Suits high-throughput projects requiring scalable reconstruction of transcriptomes from RNA-seq data.

Methodology:

Performs reference-based assembly across single or multiple RNA-seq samples using mixture statistical models for exonic feature selection, splice graph-based dynamic programming, and a weighted voting scheme for transcript selection.