proteinloc 0.1.0

Predict protein subcellular localization using protein language model embeddings.

proteinloc

Predict protein subcellular localization for prokaryotic proteins using state-of-the-art protein language model embeddings — ESM Cambrian and ProstT5.

proteinloc classifies sequences into six possible subcellular localizations:

• Cellwall
• Cytoplasmic
• CytoplasmicMembrane
• Extracellular
• OuterMembrane
• Periplasmic

Classifier artifacts (~18–23 MB each) are not bundled in the package. They are downloaded from Hugging Face Hub on first use and cached locally.

---

Installation

Using pip

pip install proteinloc

Using uv

uv tool install proteinloc

[!IMPORTANT] proteinloc requires Python ≥ 3.10 and PyTorch ≥ 2.2.
The esm package (EvolutionaryScale ESM Cambrian) requires accepting a license agreement on Hugging Face before model weights can be downloaded. Run huggingface-cli login and accept the license at https://huggingface.co/EvolutionaryScale/esmc-300m-2024-12.

---

Quick start

# Predict from a FASTA file (classifiers download automatically on first run)
proteinloc predict --fasta sequences.fasta --model esm_300

# Output as JSON
proteinloc predict --fasta sequences.fasta --model esm_600 --output-format json

# Save to a file
proteinloc predict --fasta sequences.fasta --model prost --output-format csv --output results.csv

---

CLI Reference

proteinloc predict

Embed sequences with a protein language model, then classify subcellular localization.

Option	Description	Default
--fasta PATH	Input FASTA file (required)	—
`--model [esm_300	esm_600	prost]`
`--output-format [table	json	csv]`
--output PATH	Write output to file instead of stdout	—
--device TEXT	Torch device (cpu, cuda:0, …)	auto
--weights-dir PATH	Local directory with .joblib files (overrides HF Hub)	—

proteinloc models download

Pre-fetch classifier artifacts for offline use.

# Download all models
proteinloc models download

# Download a specific model
proteinloc models download --model esm_300

proteinloc info

Show authorship, project context, and version information.

proteinloc models-list

List all available embedding models and their classifier details.

proteinloc output-formats

List available output formats and their descriptions.

---

Authorship & Context

proteinloc was developed by Juan Diego Puglia as part of a Degree Thesis in Biotechnology at Universidad ORT Uruguay.

The project aims to leverage the power of Protein Language Models (pLMs) to provide fast and accurate subcellular localization predictions for prokaryotic research.

---

Developing

To set up a local development environment:

1. Clone the repository:

   git clone https://github.com/jpuglia/proteinloc
   cd proteinloc
   

2. Install dependencies using uv:

   uv sync
   

3. Run tests:

   uv run pytest tests/ -v
   

---

License

MIT © Juan Puglia