Release 0.5.1

CHANGELOG.md

@@ -1,15 +1,35 @@
-# Changelog
+# Release notes
 
-All notable changes to this project will be documented in this file.
+Starting from version 0.5.0, this format is based on [Keep a Changelog], and this project adheres
+to [Semantic Versioning]. Full commit history is available in the
+[commit logs](https://github.com/Yoseflab/popv/commits/).
 
-The format is based on [Keep a Changelog],
-and this project adheres to [Semantic Versioning].
+For the original manuscript version 0.4.0 was used. 0.5.0 is a major rewrite of the project to scale
+to several million cells on workstations with 128 GB of RAM and enable more GPU-accelerated
+functions.
 
-## [Unreleased]
+## Version 0.5
 
-### Added
+### 0.5.2
 
-- Basic tool, preprocessing and plotting functions
+#### Fixed
+
+- Changed Louvain overclustering for Celltypist to Leiden.
+- Fixed documentation references.
+
+### 0.5.1
+
+#### Fixed
+
+- Errors in Google Colab.
+
+### 0.5.0
+
+First major release. Production-ready code.
+
+### 0.4.0
+
+Version used for the manuscript.
 
 [keep a changelog]: https://keepachangelog.com/en/1.0.0/
 [semantic versioning]: https://semver.org/spec/v2.0.0.html

README.md

@@ -88,7 +88,7 @@ our tutorials demonstrate how to generate the Natural
 Language Model used in OnClass for this user-defined ontology.
 
 ```bash
-conda create -n yourenv python=3.12
+conda create -n yourenv python=3.11
 conda activate yourenv
 pip install popv
 ```
@@ -110,4 +110,4 @@ Using a cell ontology is a requirement to run OnClass as a prediction algorithm.
 Setting ontology
 to false, will disable this step and allows running popV without using a cell ontology.
 
-[documentation]: https://popv-docs.readthedocs.io/en/latest/
+[documentation]: https://popv.readthedocs.io/en/latest/

docs/README.md

@@ -6,7 +6,6 @@
 [![Coverage](https://codecov.io/gh/YosefLab/popv/branch/main/graph/badge.svg?token=KuSsL5q3l7)](https://codecov.io/gh/YosefLab/popv)
 [![Code Style](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/python/black)
 [![Downloads](https://pepy.tech/badge/popv)](https://pepy.tech/project/popv)
-[![Docs](https://readthedocs.org/projects/popv/badge/?version=latest)](https://popv-docs.readthedocs.io/en/latest/)
 
 PopV uses popular vote of a variety of cell-type transfer tools to classify
 cell-types in a query dataset based on a test dataset. Using this variety of
@@ -97,7 +96,7 @@ package. OnClass files for annotation based on Tabula sapiens are deposited in
 `popv/resources/ontology`. We use [Cell Ontology](https://obofoundry.org/ontology/cl.html)
 as an ontology throughout our experiments. PopV will automatically look for the
 ontology in this folder. If you want to provide your user-edited ontology,
-[our tutorial](tutorials.notebooks.add_celltypes_ontology.ipynb) demonstrates how to generate the Natural
+the [custom celltype tutorial](tutorials/notebooks/add_celltypes_ontology) demonstrates how to generate the Natural
 Language Model used in OnClass for this user-defined ontology.
 
 ```bash
@@ -110,7 +109,7 @@ pip install popv
 
 We provide an example notebook in Google Colab:
 
-- [Tutorial demonstrating use of Tabula sapiens as a reference](tutorials.notebooks.tabula_sapiens_tutorial.ipynb)
+- [Tutorial demonstrating use of Tabula sapiens as a reference](tutorials/notebooks/tabula_sapiens_tutorial)
 
 This notebook will guide you through annotating a dataset based on the annotated
 [Tabula sapiens reference](https://tabula-sapiens-portal.ds.czbiohub.org) and

docs/conf.py

@@ -138,7 +138,6 @@
 html_theme_options = {
     "repository_url": repository_url,
     "use_repository_button": True,
-    "logo_only": True,
     "show_toc_level": 1,
     "launch_buttons": {"colab_url": "https://colab.research.google.com"},
     "path_to_docs": "docs/",

popv/__init__.py

@@ -5,8 +5,8 @@
 
 if sys.version_info[:2] != (3, 10):
     warnings.warn(
-        "Pretrained models on huggingface are trained with Python 3.10. "
-        "Detected Python {sys.version.split()[0]} will not load these models.",
+        "Pretrained models on huggingface are trained with Python 3.11. "
+        f"Detected Python {sys.version.split()[0]} will not load these models.",
         UserWarning,
         stacklevel=3,
     )

popv/algorithms/_celltypist.py

@@ -92,10 +92,9 @@ def predict(self, adata):
             ].cat.categories[mode(neighbor_values, axis=1).mode.flatten()]
             over_clustering = adata.obs.loc[adata.obs["_predict_cells"] == "relabel", "over_clustering"]
         else:
-            flavor = "rapids" if settings.cuml else "vtraag"
             transformer = "rapids" if settings.cuml else None
             sc.pp.neighbors(adata, n_neighbors=15, use_rep="X_pca", transformer=transformer)
-            sc.tl.louvain(adata, resolution=25.0, key_added="over_clustering", flavor=flavor)
+            sc.tl.leiden(adata, resolution=25.0, key_added="over_clustering")
             over_clustering = adata.obs.loc[adata.obs["_predict_cells"] == "relabel", "over_clustering"]
 
         if adata.uns["_prediction_mode"] == "retrain":

popv/hub/_metadata.py

@@ -118,13 +118,18 @@ class HubModelCardHelper:
     metrics_report
         A dictionary containing the metrics report for the model.
 
+    Attributes
+    ----------
+    model_card : ModelCard
+        Stores the model card.
+
     Notes
     -----
     It is not required to use this class to create a `ModelCard`. But this helps you do so in a way
-    that is consistent with most other `popV` hub models. You can think of this as a
-    template. The actual template string used can be found in ``popv.hub.template``. The resulting
+    that is consistent with other `popV` hub models. You can think of this as a
+    template. The resulting
     huggingface :class:`~huggingface_hub.ModelCard` can be accessed via the
-    :meth:`~popv.hub.HubModelCardHelper.model_card` property.
+    :attr:`~popv.hub.HubModelCardHelper.model_card` property.
     """
 
     license_info: str

pyproject.toml

@@ -10,7 +10,7 @@ exclude = ["resources/*", "tests/*", "dataset/*"]
 
 [project]
 name = "popV"
-version = "0.5.1post1"
+version = "0.5.2"
 description = "Consensus prediction of cell type labels with popV"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -51,7 +51,7 @@ dependencies = [
     "harmony-pytorch",
     "huggingface",
     "xgboost",
-    "louvain",
+    "leidenalg",
     "cellxgene_census",
 ]