Translation Efficiency - Usage Guide

Overview

Quantify translation efficiency (TE = ribosome occupancy / mRNA abundance) and test for differential TE between conditions. The key decisions are counting both assays over the CDS, using count-based GLMs rather than ratio tests, and distinguishing genuine translational control from buffering.

Prerequisites

# R (differential TE)
BiocManager::install(c("riborex", "xtail", "anota2seq", "DESeq2"))

# Python (quick ranking screen)
pip install pandas numpy statsmodels

Quick Start

Tell your AI agent what you want to do: - "Calculate translation efficiency from my Ribo-seq and RNA-seq CDS counts" - "Find genes with differential TE with riborex" - "Tell translational control from buffering with anota2seq" - "Run a DESeq2 interaction model for TE"

Example Prompts

Calculate TE

"Compute per-gene log2 TE over the CDS for ranking"

"Why do my long-UTR genes have systematically low TE?"

Differential TE

"Run riborex to find genes with significantly changed TE"

"Use anota2seq to classify genes as translation, buffering, or abundance"

"Set up a DESeq2 condition-by-assay interaction and pick the right coefficient"

Interpretation

"Which genes are translationally activated versus buffered?"

"Is this TE change real translational control or secondary to a uORF?"

What the Agent Will Do

1. Load matched Ribo-seq and RNA-seq CDS count matrices
2. For ranking, compute per-gene log2 TE with a pseudocount
3. For testing, run a count-based GLM (riborex/Xtail/anota2seq/DESeq2 interaction)
4. With anota2seq, classify each gene's mode of regulation
5. Filter by adjusted p-value and check uORF/isoform confounders

Tips

• Count both over the CDS - full-transcript RNA against CDS RPF biases long-UTR genes
• Ratios rank, GLMs test - never t-test log-TE for inference
• Buffering looks like control - only anota2seq separates them
• Do not hardcode the DESeq2 interaction name - pick it from resultsNames(dds)
• Per-assay size factors - assume median TE unchanged; use spike-ins for global shifts
• Trim codons - exclude first ~15 / last ~5 codons from the RPF count
• Replicates - count GLMs and anota2seq RVM want >=3 per condition per assay; n=2 is illustrative only

Related Skills

• ribosome-periodicity - Calibrate P-site offsets for CDS counts
• orf-detection - Rule out uORF-driven TE changes
• rna-quantification/featurecounts-counting - Matched RNA-seq CDS counts
• differential-expression/deseq2-basics - Count-based DE foundations