Sequgio

Sequgio estimates isoform-level expression from RNA-seq data by modeling isoform-specific non-uniform read distributions to improve accuracy of isoform abundance quantification.

Key Features:

• Non-uniform read distribution modeling: Incorporates a joint statistical model that accounts for isoform-specific non-uniform read distributions within transcripts from RNA-seq data.
• Statistical regularization: Applies a smoothing penalty to control complexity arising from large numbers of potential splice variants and their read distribution parameters.
• Cross-sample information integration: Leverages information across multiple samples from the same genomic region to improve identifiability and robustness of estimates.
• Iterated-weighted least-squares algorithm: Uses an iterated-weighted least-squares computational procedure for efficient parameter estimation.
• Empirical validation: Evaluated on simulated datasets and real RNA-seq datasets with validation by reverse transcription-polymerase chain reaction (RT-PCR).
• Implementation: Provided as an R package implementation.

Scientific Applications:

• Isoform-level quantification: Generates more accurate estimates of isoform abundance from RNA-seq for isoform-resolved expression studies.
• Alternative splicing analysis: Supports analyses of alternative splicing events by accounting for isoform-specific read distribution differences.
• Transcriptome complexity studies: Facilitates investigation of transcriptome complexity by improving precision of isoform expression measurements.

Methodology:

Uses a joint statistical model for isoform-specific non-uniform read distributions with a smoothing-penalty regularization, integrates information across samples for identifiability, and estimates parameters via an iterated-weighted least-squares algorithm; evaluated using simulated datasets and RT-PCR.