DeepShape

DeepShape estimates isoform-level ribosome abundance and distribution from ribosome profiling (Ribo-seq) data using a deep learning model that operates without RNA-seq input.

Key Features:

• Isoform-Level Precision: Assigns multiple-mapped Ribo-seq reads to similar isoforms to enable isoform-specific ribosome abundance and profile estimation.
• Deep Learning Integration: Employs a deep learning model to simultaneously estimate ribosome abundance and compute ribosome profiles for each isoform.
• RNA-seq Independence: Estimates transcript-level ribosome metrics without requiring RNA-seq data for transcript abundance.
• Codon Residence Index (CRI): Introduces CRI as a metric that measures the relative speed at which ribosomes traverse a codon compared to its synonymous counterparts.

Scientific Applications:

• PC3 cell PP242 analysis: Applied to Ribo-seq data from PC3 human prostate cancer cells with and without PP242 treatment to study translation regulation at the isoform level.
• Isoform- and codon-level translation analysis: Revealed distinct translational efficiency and regulatory patterns among isoforms of genes involved in cell invasion and metastasis affected by PP242, with CRI indicating consistent codon-level translational patterns.

Methodology:

Uses a deep learning model to assign multi-mapped Ribo-seq reads and to estimate isoform-level ribosome abundance and profiles without RNA-seq; computes the Codon Residence Index (CRI) to quantify relative ribosome traversal speed across synonymous codons.