Strawberry

Strawberry reconstructs and quantifies transcripts ab initio from RNA-seq data using genome alignment guidance for transcriptome analysis.

Key Features:

• Assembly module: Constructs splicing graphs from aligned RNA-seq reads.
• Network flow optimization: Uses network flow algorithms to identify the most probable transcripts.
• Quantification module: Applies a latent class model to allocate read counts from splicing-graph nodes to specific transcripts.
• Bias correction: Employs an Expectation-Maximization (EM) algorithm to correct sequencing biases during quantification.
• Unified data graph: Uses a unified data graph structure across assembly and quantification to handle large datasets efficiently.
• Ab initio operation: Operates independently of gene annotation while leveraging genome alignment guidance.
• Benchmarking: Validated on simulations and real data with superior assembly and quantification accuracy compared to Cufflinks and StringTie.
• Nanostring concordance: Estimated transcript expression levels exhibit high correlation with Nanostring probe counts.
• Implementation: Implemented in C++14.

Scientific Applications:

• Transcriptome assembly: Ab initio reconstruction of transcripts from RNA-seq alignments.
• Transcript quantification: Estimation of transcript abundances with explicit sequencing bias correction.
• Method benchmarking: Comparative evaluation of transcript assembly and quantification tools using simulations, real data, and Nanostring probes.
• Large-scale RNA-seq analysis: Integrated assembly and quantification for handling large RNA-seq datasets.

Methodology:

Constructs splicing graphs from aligned RNA-seq reads, applies network flow algorithms to assemble transcripts, uses a latent class model with an Expectation-Maximization algorithm to allocate reads and correct sequencing biases, and implements these methods via a unified data graph in C++14.