Fix TestPyPI publish failure: remove git URL from metl optional dependency

.github/workflows/test.yml

@@ -14,6 +14,9 @@ concurrency:
   group: tests-${{ github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: read
+
 jobs:
   lint:
     name: Lint & format (pre-commit)

CHANGELOG.md

@@ -19,10 +19,10 @@ truth that drives publishing).
 
 ### Added
 - METL 1D protein embeddings via optional `metl-pretrained` backend: install with
-  `pip install pepe-cli[metl]` and use model identifiers such as `metl-g-20m-1d`
+  `pip install git+https://github.com/gitter-lab/metl-pretrained.git` and use model identifiers such as `metl-g-20m-1d`
   (and other `metl-*-1d` names). Dispatch lives in `model_selecter.py`; embedding
   is handled by `METLEmbedder` with `METLDataset` tokenization.
-- `[metl]` and `[esm]` optional dependency extras in `pyproject.toml` and
+- `[esm]` optional dependency extra in `pyproject.toml` and
   `setup.py`.
 - Typed errors `METLPackageRequiredError` and `METL3DNotSupportedError` when
   METL is requested without the extra or when a 3D METL model id is used.

README.md

@@ -32,13 +32,7 @@ PEPE (Pipeline for Easy Protein Embedding) is a tool for extracting embeddings a
     pip install git+https://github.com/Biohub/transformers.git@main
     ```
 
-3. *(Optional)* For METL 1D embedding models (e.g. `metl-g-20m-1d`), install the optional backend:
-
-    ```sh
-    pip install pepe-cli[metl]
-    ```
-
-    The underlying `metl-pretrained` package is not on PyPI. PEPE's `[metl]` extra installs it from GitHub; you can also install it directly:
+3. *(Optional)* For METL 1D embedding models (e.g. `metl-g-20m-1d`), install the backend directly from GitHub (it is not on PyPI):
 
     ```sh
     pip install git+https://github.com/gitter-lab/metl-pretrained.git
@@ -233,7 +227,7 @@ results = pepe.embed(
         - biohub/ESMC-300M
         - biohub/ESMC-600M
         - biohub/ESMC-6B
-    - METL 1D models (requires `[metl]` extra; see Quick start)
+    - METL 1D models (requires `metl-pretrained` from GitHub; see Quick start)
         - `metl-g-20m-1d` and other `metl-*-1d` identifiers supported by [metl-pretrained](https://github.com/gitter-lab/metl-pretrained)
     - Custom Hugging Face models
         - Any compatible model from Hugging Face Hub: `username/model-name`
@@ -250,7 +244,7 @@ results = pepe.embed(
 - **`--model_name`** (str): Name of model or link to model. Choose from [List of supported models](../README.md#list-of-supported-models) or use custom models:
   - ESM models: `esm2_t33_650M_UR50D`
   - ESMC models: `biohub/ESMC-300M` (requires Biohub transformers fork; see Quick start)
-  - METL 1D models: `metl-g-20m-1d` (requires `[metl]` extra; see Quick start)
+  - METL 1D models: `metl-g-20m-1d` (requires `metl-pretrained` from GitHub; see Quick start)
   - Hugging Face models: `username/model-name`
   - Custom PyTorch models: `/path/to/model.pt` or `/path/to/model_directory/`
   - Local HF models: `/path/to/local_hf_directory/`

pyproject.toml

@@ -35,7 +35,6 @@ dependencies = [
 ]
 
 [project.optional-dependencies]
-metl = ["metl-pretrained @ git+https://github.com/gitter-lab/metl-pretrained.git@52358614c4b412e81e19e300485e8b85123bd903"]
 esm = ["fair-esm"]
 # Developer toolchain: `pip install -e .[dev]` bootstraps everything needed to
 # run the hooks and the local test suite.

setup.py

@@ -57,9 +57,6 @@ def read_readme():
         "rjieba",
     ],
     extras_require={
-        "metl": [
-            "metl-pretrained @ git+https://github.com/gitter-lab/metl-pretrained.git@52358614c4b412e81e19e300485e8b85123bd903"
-        ],
         "esm": ["fair-esm"],
     },
     entry_points={

src/pepe/embedders/metl_embedder.py

@@ -16,7 +16,6 @@ def _import_metl():
     except ImportError as e:
         raise METLPackageRequiredError(
             "METL models require metl-pretrained. Install with: "
-            "pip install 'pepe-cli[metl]' or "
             "pip install git+https://github.com/gitter-lab/metl-pretrained.git"
         ) from e