Publish to PyPI

.github/workflows/build-and-test.yml

@@ -11,7 +11,7 @@ jobs:
     strategy:
       matrix:
         os: [ubuntu-latest]
-        python-version: [3.9]
+        python-version: [3.8, "3.11"]
 
     runs-on: ${{ matrix.os }}
 
@@ -51,3 +51,4 @@ jobs:
         run: |
           cd main
           make test
+          make notebooks

.github/workflows/pypi-publish.yml

@@ -0,0 +1,36 @@
+# This workflow will upload a Python Package using Twine when a release is created
+# For more information see: https://help.github.com/en/actions/language-and-framework-guides/using-python-with-github-actions#publishing-to-package-registries
+
+# This workflow uses actions that are not certified by GitHub.
+# They are provided by a third-party and are governed by
+# separate terms of service, privacy policy, and support
+# documentation.
+
+name: Upload Python Package
+
+on:
+  release:
+    types: [published]
+
+jobs:
+  deploy:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Set up Python
+      uses: actions/setup-python@v2
+      with:
+        python-version: '3.9'
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install build
+    - name: Build package
+      run: python -m build
+    - name: Publish package
+      uses: pypa/gh-action-pypi-publish@27b31702a0e7fc50959f5ad993c78deac1bdfc29
+      with:
+        user: __token__
+        password: ${{ secrets.PYPI_TOKEN }}

Makefile

@@ -24,7 +24,7 @@ docs:
 notebooks:
 	mkdir -p notebooks/_ignore
 	for nb in notebooks/*.ipynb; do \
-		jupyter nbconvert --to notebook --execute "$$nb" --output notebooks/_ignore/"$$(basename $$nb)"; \
+		jupyter nbconvert --to notebook --execute "$$nb" --output _ignore/"$$(basename $$nb)"; \
 	done
 
 .PHONY: install test notebooks format lint docs

README.md

@@ -4,7 +4,11 @@ Neural NETworks for antibody Affinity Maturation.
 
 ## pip installation
 
-TODO
+Netam is available on PyPI, and works with Python 3.8 through 3.11.
+
+```
+pip install netam
+```
 
 This will allow you to use the models.
 
@@ -55,4 +59,4 @@ If you are running one of the experiment repos, such as:
 * [thrifty-experiments-1](https://github.com/matsengrp/thrifty-experiments-1/)
 * [dnsm-experiments-1](https://github.com/matsengrp/dnsm-experiments-1/)
 
-you will want to visit those repos and follow the installation instructions there.
+you will want to visit those repos and follow the installation instructions there.

netam/multihit.py

@@ -14,7 +14,7 @@
 from torch.utils.data import Dataset
 from tqdm import tqdm
 import pandas as pd
-from typing import Sequence
+from typing import Sequence, List, Tuple
 
 from netam.molevol import (
     reshape_for_codons,
@@ -30,8 +30,8 @@
 
 
 def _trim_to_codon_boundary_and_max_len(
-    seqs: list[Sequence], max_len: int = None
-) -> list[Sequence]:
+    seqs: List[Sequence], max_len: int = None
+) -> List[Sequence]:
     """Trims sequences to codon boundary and maximum length.
 
     No assumption is made about the data of a sequence, other than that it is
@@ -55,7 +55,7 @@ def _observed_hit_classes(parents: Sequence[str], children: Sequence[str]):
         children (Sequence[str]): A list of the corresponding child sequences.
 
     Returns:
-        list[torch.Tensor]: A list of tensors, each containing the observed
+        List[torch.Tensor]: A list of tensors, each containing the observed
             hit classes for each codon in the parent sequence. At any codon position
             where the parent or child sequence contains an N, the corresponding tensor
             element will be -100.
@@ -96,8 +96,8 @@ def __init__(
         self,
         nt_parents: Sequence[str],
         nt_children: Sequence[str],
-        nt_ratess: Sequence[list[float]],
-        nt_cspss: Sequence[list[list[float]]],
+        nt_ratess: Sequence[List[float]],
+        nt_cspss: Sequence[List[List[float]]],
         branch_length_multiplier: float = 1.0,
     ):
         trimmed_parents = _trim_to_codon_boundary_and_max_len(nt_parents)
@@ -452,7 +452,7 @@ def hit_class_dataset_from_pcp_df(
 
 def train_test_datasets_of_pcp_df(
     pcp_df: pd.DataFrame, train_frac: float = 0.8, branch_length_multiplier: float = 1.0
-) -> tuple[HitClassDataset, HitClassDataset]:
+) -> Tuple[HitClassDataset, HitClassDataset]:
     """Splits a pcp_df prepared by `prepare_pcp_df` into a training and testing
     HitClassDataset."""
     nt_parents = pcp_df["parent"].reset_index(drop=True)

netam/pretrained.py

@@ -11,22 +11,24 @@
 
 from netam.framework import load_crepe
 
+# This throws a deprecation warning. It could also be done by looking at
+# __file__, or by using importlib.resources.
 PRETRAINED_DIR = pkg_resources.resource_filename(__name__, "_pretrained")
 
 PACKAGE_LOCATIONS_AND_CONTENTS = (
     # Order of entries:
     # * Local file name
     # * Remote URL
-    # * Directory in which the models appear after extraction
+    # * Directory in which the models appear after extraction (must match path determined by archive)
     # * List of models in the package
     [
-        "thrifty-1.0.zip",
-        "https://github.com/matsengrp/thrifty-models/archive/refs/heads/release/1.0.zip",
-        "thrifty-models-release-1.0/models",
+        "thrifty-0.2.0.zip",
+        "https://github.com/matsengrp/thrifty-models/archive/refs/tags/v0.2.0.zip",
+        "thrifty-models-0.2.0/models",
         [
-            "ThriftyHumV1.0-20",
-            "ThriftyHumV1.0-45",
-            "ThriftyHumV1.0-59",
+            "ThriftyHumV0.2-20",
+            "ThriftyHumV0.2-45",
+            "ThriftyHumV0.2-59",
         ],
     ],
 )
@@ -39,7 +41,7 @@
     LOCAL_TO_REMOTE[local_file] = remote
 
     for model in models:
-        MODEL_TO_LOCAL[model] = local_file
+        MODEL_TO_LOCAL[model] = (local_file, models_dir)
 
 
 def local_path_for_model(model_name: str):
@@ -50,7 +52,7 @@ def local_path_for_model(model_name: str):
 
     os.makedirs(PRETRAINED_DIR, exist_ok=True)
 
-    local_package = MODEL_TO_LOCAL[model_name]
+    local_package, models_dir = MODEL_TO_LOCAL[model_name]
     local_package_path = os.path.join(PRETRAINED_DIR, local_package)
 
     if not os.path.exists(local_package_path):
@@ -62,14 +64,16 @@ def local_path_for_model(model_name: str):
             f.write(response.content)
         if local_package.endswith(".zip"):
             with zipfile.ZipFile(local_package_path, "r") as zip_ref:
-                zip_ref.extractall(PRETRAINED_DIR)
+                zip_ref.extractall(path=PRETRAINED_DIR)
         else:
             raise ValueError(f"Unknown file type for {local_package}")
+    else:
+        print(f"Using cached models: {local_package_path}")
 
     local_crepe_path = os.path.join(PRETRAINED_DIR, models_dir, model_name)
 
     if not os.path.exists(local_crepe_path + ".yml"):
-        raise ValueError(f"Model {model_name} not found in pre-trained models.")
+        raise ValueError(f"Model {local_crepe_path} not found in pre-trained models.")
     if not os.path.exists(local_crepe_path + ".pth"):
         raise ValueError(f"Model {model_name} missing model weights.")
 

pyproject.toml

@@ -1,3 +1,5 @@
 [build-system]
-requires = ["setuptools>=64", "wheel"]
+requires = ["setuptools>=64", "wheel", "setuptools-scm>=8"]
 build-backend = "setuptools.build_meta"
+
+[tool.setuptools_scm]

requirements.txt

@@ -2,7 +2,10 @@ black
 docformatter
 fire
 nbconvert
+ipython
+ipykernel
 pytest
 snakemake
 tensorboardX
 typing_extensions
+seaborn

setup.py

@@ -1,15 +1,17 @@
 from setuptools import setup, find_packages
 
+# Version is determined by setuptools-scm in pyproject.toml according to git
+# tag numbering.
 setup(
     name="netam",
-    version="0.1.0",
     url="https://github.com/matsengrp/netam.git",
     author="Matsen Group",
     author_email="ematsen@gmail.com",
     description="Neural network models for antibody affinity maturation",
     long_description=open("README.md").read(),
     long_description_content_type="text/markdown",
     packages=find_packages(),
+    python_requires=">=3.8,<3.12",
     install_requires=[
         "biopython",
         "natsort",

tests/test_molevol.py

@@ -144,7 +144,7 @@ def iterative_aaprob_of_mut_and_sub(parent_codon, mut_probs, csps):
 
 
 def test_aaprob_of_mut_and_sub():
-    crepe = pretrained.load("ThriftyHumV1.0-45")
+    crepe = pretrained.load("ThriftyHumV0.2-45")
     [rates], [subs] = crepe([parent_nt_seq])
     mut_probs = 1.0 - torch.exp(-rates.squeeze().clone().detach())
     parent_codon = parent_nt_seq[0:3]