MultiTrans

MultiTrans formulates transcriptome assembly as a mixed integer linear programming problem to extract expressed transcript paths from splicing (assembly) graphs built from RNA-seq reads.

Key Features:

• Mixed Integer Linear Programming (MILP): Solves the path extraction problem on splicing/assembly graphs using MILP.
• Path extraction on splicing/assembly graphs: Models transcripts as paths through splicing (assembly) graphs to represent splicing variants.
• Vertex- and edge-level coverage constraints: Incorporates coverage information at both vertex and edge levels to align assembled transcripts with observed read coverage.
• Path-number limitation: Controls the number of paths extracted from the graph to manage computational complexity and reconstruction accuracy.
• Paired-end read integration: Integrates paired-end information to resolve ambiguities in the assembly graph.
• Robustness to transcriptomic complexity: Addresses splicing variants, varying expression levels, uneven coverage, and sequencing errors during assembly.

Scientific Applications:

• Transcript reconstruction from RNA-seq: Reconstructs expressed transcripts from RNA-seq reads using splicing/assembly graphs.
• Splice variant resolution: Enables reconstruction and differentiation of alternative splicing isoforms.
• Benchmarking and comparative assembly evaluation: Produced more accurate transcripts than TransLiG and rnaSPAdes when evaluated on identical splicing/assembly graphs in benchmark studies.
• Gene expression and regulation analysis: Provides reconstructed transcript structures to support analyses of genes, proteins, and cellular functions.

Methodology:

Formulates transcriptome assembly as a path extraction task on splicing/assembly graphs and solves it using mixed integer linear programming while incorporating vertex- and edge-level coverage constraints, path-number limits, and paired-end read information.