seq2ribo¶

Predict ribosome profiling signals, translation efficiency, or protein expression from mRNA sequence.

Paper: bioRxiv 2026
Upstream: https://github.com/Kingsford-Group/seq2ribo
License: CMU Academic/Non-Commercial Research Use Only (commercial use prohibited)
Device: GPU only (CUDA required, auto-skipped under --profile cpu)

What it does¶

seq2ribo uses a Mamba state-space model to predict ribosome profiling data from mRNA sequence alone. It supports three prediction tasks:

riboseq: per-position ribosome density profile
te: translation efficiency (scalar per transcript)
protein: protein expression level (scalar per transcript)

Models are available for 4 human cell lines: HEK293, LCL, RPE, iPSC.

Input format¶

FASTA file of mRNA coding sequences (CDS only, or with UTRs if using --use_utr).

Example (tests/data/seq2ribo_test.fa):

>test_cds_1
AUGGCCAAAGCUACCUUGAAGCGAGCACUGGAGAGCUUUGCCAAGGCCUUGAAAGAGAUG
AAAGCCACCAAAGAAUGGAAUGCCGUGACCAAAGCUGCCAGCAUCCUGAAAGAGAAAGCCC
UGGAUUAG

Output format¶

JSON file with per-sequence predictions:

[
  {
    "id": "test_cds_1",
    "prediction": 2.2616412272516313
  }
]

For the riboseq task, prediction is an array of per-position values instead of a scalar.

Run with Docker¶

See the Direct Docker guide for the shared docker run recipe (UID, HOME, USER env vars, and GPU flag). Below are the model-specific parts.

docker run --rm \
  --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=all \
  -v /path/to/input.fa:/data/input.fa \
  -v /path/to/output:/out \
  ghcr.io/ericmalekos/rnazoo-seq2ribo:latest \
  /opt/conda/envs/seq2ribo/bin/python /opt/seq2ribo/scripts/run_inference.py \
    --task te \
    --cell-line hek293 \
    --fasta /data/input.fa \
    --output /out/seq2ribo_output.json

Note: on systems with snap-installed Docker, use --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=all instead of --gpus all.

Run with Nextflow¶

nextflow run main.nf -profile docker,gpu \
  --seq2ribo_input /path/to/input.fa \
  --seq2ribo_task te \
  --seq2ribo_cell_line hek293

Only models with input provided will run — no ignore flags needed.

Parameters¶

Parameter	Default	Description
`--seq2ribo_task`	`te`	Prediction task: `riboseq`, `te`, or `protein`
`--seq2ribo_cell_line`	`hek293`	Cell line: `hek293`, `lcl`, `rpe`, or `ipsc`
`--seq2ribo_use_utr`	`false`	Include UTR regions in prediction

Example output¶

[
  {
    "id": "test_cds_1",
    "prediction": 2.2616412272516313
  }
]

A TE value of ~2.26 indicates the transcript is predicted to be translated at about 2^2.26 = ~4.8x the average efficiency in HEK293 cells.