TranslationAI¶

Predict translation initiation sites (TIS), translation termination sites (TTS), and ORFs from mRNA sequence.

Paper: NAR 2025
Upstream: https://github.com/rnasys/TranslationAI
License: AGPL-3.0 (source code) + CC BY-NC 4.0 (model weights, non-commercial only)
Device: CPU or GPU. Two image variants:
- rnazoo-translationai:latest — CUDA-enabled (TF 1.15 GPU build, default with -profile gpu). Works on any GPU; older binary kernels PTX-JIT for compute capabilities >7.0 (Turing/Ampere/Ada/Hopper) on first kernel launch.
- rnazoo-translationai-cpu:latest — CPU-only (smaller, used with -profile cpu)

What it does¶

TranslationAI uses deep learning to identify translation initiation sites (TIS), translation termination sites (TTS), and complete ORFs in mRNA sequences. It scans each input mRNA and outputs site-level probability scores for where translation starts and stops, then assembles predicted ORFs from TIS/TTS pairs.

Input format¶

FASTA file of mRNA sequences (DNA alphabet):

>chr3:28283123-28361264(+)(CMC1)(241, 568,)
GAGGGAACGGGTCCTGGCGGTGCTTTGCAAAGGGCCCGTGTTTCTGTTG...
>chr9:98997588-99064434(-)(HSD17B3)(48, 978,)
TACACAGAGAGCCACGGCCAGGGCTGAAACAGTCTGTTGAGTGCAGCCAT...

Headers can be in any format. The optional (start, stop,) notation in the example above is used by the upstream tool for validation but is not required.

Output format¶

Three text files per input:

*_predTIS_*.txt -- predicted TIS positions and probabilities:
```
>chr3:28283123-28361264(+)(CMC1)(241, 568,)  241,0.8824475854635239
```
Format: header<TAB>position,probability
*_predTTS_*.txt -- predicted TTS positions and probabilities (same format)
*_predORFs_*.txt -- predicted ORFs (TIS-TTS pairs with scores)

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 \
  -v /path/to/input.fa:/data/input.fa \
  -v /path/to/output:/out \
  ghcr.io/ericmalekos/rnazoo-translationai:latest \
  bash -c "cp /data/input.fa /tmp/input.fa && \
    translationai -I /tmp/input.fa -t 0.5,0.5 && \
    cp /tmp/input_pred*.txt /out/"

Run with Nextflow¶

nextflow run main.nf -profile docker,cpu \
  --translationai_input /path/to/input.fa

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

Parameters¶

Parameter	Default	Description
`--translationai_tis_threshold`	`0.5`	TIS probability threshold (0-1)
`--translationai_tts_threshold`	`0.5`	TTS probability threshold (0-1)

Example output¶

TIS predictions:

>chr3:28283123-28361264(+)(CMC1)(241, 568,) 241,0.8824475854635239
>chr9:98997588-99064434(-)(HSD17B3)(48, 978,)   48,0.9925230294466019
>chr1:153175905-153177601(+)(LELP1)(123, 417,)  123,0.9739317893981934

Each line shows the predicted TIS position and its probability score. Higher scores indicate more confident predictions. The threshold parameters control which predictions are reported.