Merge pull request #15 from mvinyard/demo

Demo

README.md

@@ -9,7 +9,7 @@ Create [`PyTorch Datasets`](https://pytorch.org/tutorials/beginner/basics/data_t
 
 ## Installation
 
-Install from PYPI (current version: **[`0.0.23`](https://pypi.org/project/torch-adata/)**):
+Install from PYPI (current version: **[`0.0.24`](https://pypi.org/project/torch-adata/)**):
 ```BASH
 pip install torch-adata
 ```

setup.py

@@ -2,7 +2,7 @@
 __module_name__ = "setup.py"
 __doc__ = """PYPI package distribution setup."""
 __author__ = ", ".join(["Michael E. Vinyard"])
-__email__ = ", ".join(["[email protected]"])
+__email__ = ", ".join(["[email protected]"])
 
 
 # -- import packages: ----------------------------------------------------------
@@ -15,10 +15,10 @@
 # -- run setup: ----------------------------------------------------------------
 setuptools.setup(
     name="torch-adata",
-    version="0.0.23",
+    version="0.0.24",
     python_requires=">3.9.0",
     author="Michael E. Vinyard",
-    author_email="mvinyard@g.harvard.edu",
+    author_email="mvinyard[email protected]",
     url=None,
     long_description=open("README.md", encoding="utf-8").read(),
     long_description_content_type="text/markdown",
@@ -28,11 +28,16 @@
         "anndata>=0.9.1",
         "licorice_font>=0.0.3",
         "lightning>=2.0.1",
-        "torch>=2.0",        
+        "torch>=2.0",
+#         "numpy==1.23",
+        "scanpy==1.9.3",
+        "scikit-learn==1.2.2",
+        "webfiles",
+	"vinplots",
     ],
     classifiers=[
         "Development Status :: 2 - Pre-Alpha",
-        "Programming Language :: Python :: 3.7",
+        "Programming Language :: Python :: 3.9",
         "Intended Audience :: Science/Research",
         "Topic :: Scientific/Engineering :: Bio-Informatics",
     ],

torch_adata/__init__.py

@@ -6,7 +6,7 @@
 
 
 # -- specify package version: --------------------------------------------------
-__version__ = "0.0.23"
+__version__ = "0.0.24"
 
 
 # -- import modules: -----------------------------------------------------------
@@ -18,4 +18,4 @@
 # -- import API-hidden core modules (for dev): ---------------------------------
 from ._core import _core_ancilliary as _core
 from . import _utils
-from . import _demo
+from . import _demo

torch_adata/_demo/__init__.py

@@ -0,0 +1,5 @@
+
+from ._plot_umap import PlotUMAP
+from ._pbmc_3k import PBMC3k
+from ._larry import LARRY
+from ._demo_data import DemoData

torch_adata/_demo/_demo_data.py

@@ -0,0 +1,26 @@
+
+
+from .. import _utils as utils
+from ._pbmc_3k import PBMC3k
+from ._larry import LARRY
+
+
+class DemoData(utils.ABCParse):
+    def __init__(self):
+        """"""
+
+    @property
+    def PBMC3k(self):
+        if not hasattr(self, "_pbmcs"):
+            self._pbmcs = PBMC3k()
+        else:
+            self._pbmcs.plot()
+        return self._pbmcs.adata
+
+    @property
+    def LARRY(self):
+        if not hasattr(self, "_larry"):
+            self._larry = LARRY()
+        else:
+            self._larry.plot()
+        return self._larry.adata

torch_adata/_demo/_larry.py

@@ -0,0 +1,56 @@
+
+import vinplots
+import anndata
+import os
+import web_files
+
+from .. import _utils as utils
+from ._plot_umap import PlotUMAP
+
+class LARRY(utils.ABCParse):
+    _HTTP_ADDRESS = "https://figshare.com/ndownloader/files/38171943"
+
+    def __init__(self, filename="adata.LARRY.h5ad", data_dir="data", silent=False):
+        self._config(locals())
+
+    def _config(self, kwargs):
+        """"""
+
+        self.__parse__(kwargs, public=[None])
+
+        self._INFO = utils.InfoMessage()
+
+        if not os.path.exists(self._data_dir):
+            os.mkdir(self._data_dir)
+
+    def _download(self):
+        msg = "Downloading {:.<10}...".format(self._filename)
+        self._INFO(msg, end=" ")
+        f = web_files.WebFile(
+            http_address=self._HTTP_ADDRESS,
+            local_path=self.local_path,
+        )
+        f.download()
+        print("Done.")
+
+    @property
+    def local_path(self):
+        return os.path.join(self._data_dir, self._filename)
+
+    @property
+    def adata(self):
+        if not hasattr(self, "_adata"):
+            if not os.path.exists(self.local_path):
+                self._download()
+            self._adata = anndata.read_h5ad(self.local_path)
+            self._adata.uns["cmap"] = vinplots.colors.LARRY_in_vitro
+            if not self._silent:
+                print(self._adata)
+                self.plot()
+        return self._adata
+
+    def plot(self):
+        umap_plot = PlotUMAP(
+            self._adata, title="LARRY hematopoiesis"
+        )
+        umap_plot(groupby="Cell type annotation")

torch_adata/_demo/_pbmc_3k.py

@@ -0,0 +1,39 @@
+
+# -- import packages: --------------------------------------------------------
+import scanpy as sc
+import pandas as pd
+import vinplots
+
+
+from .. import _utils as utils
+from ._plot_umap import PlotUMAP
+
+class PBMC3k(utils.ABCParse):
+    def __init__(self, silent=False):
+        self.__parse__(locals(), public=[None])
+
+    def _configure(self):
+        _adata = sc.datasets.pbmc3k_processed()
+        cell_type_df = pd.DataFrame(_adata.obs["louvain"].unique()).reset_index()
+        cell_type_df.columns = ["cell_type_idx", "louvain"]
+        df_obs = _adata.obs.merge(cell_type_df, on="louvain", how="left")
+        df_obs.index = df_obs.index.astype(str)
+        _adata.obs = df_obs
+        _adata.uns["cmap"] = vinplots.colors.pbmc3k
+        return _adata
+
+    @property
+    def adata(self):
+        if not hasattr(self, "_adata"):
+            self._adata = self._configure()
+            if not self._silent:
+                print(self._adata)
+                self.plot()
+        return self._adata
+
+    def plot(self):
+        umap_plot = PlotUMAP(
+            self._adata, title="10x PBMCs (~3k cells)"
+        )
+        umap_plot(groupby="louvain")
+

torch_adata/_demo/_plot_umap.py

@@ -0,0 +1,65 @@
+
+from .. import _utils as utils
+
+import vinplots
+
+
+class PlotUMAP(utils.ABCParse):
+    def __init__(
+        self,
+        adata,
+        title="10x PBMCs (~3k cells)",
+        wspace=0.1,
+        rm_ticks=True,
+        spines_to_delete=["top", "bottom", "right", "left"],
+        labels=["UMAP-1", "UMAP-2"],
+        use_key="X_umap",
+        cmap_key="cmap",
+        alpha=0.65,
+    ):
+        self.__parse__(locals(), public=["adata"])
+
+    def __config__(self):
+        qiuck_plot_params = utils.FunctionInspector(func=vinplots.quick_plot)
+
+        self.fig, self.axes = vinplots.quick_plot(
+            nplots=1, ncols=1, **qiuck_plot_params(self._PARAMS)
+        )
+
+    @property
+    def CMAP(self):
+        return self.adata.uns[self._cmap_key]
+
+    def _plot_grouped(self):
+        for en, (group, group_df) in enumerate(self.adata.obs.groupby(self.groupby)):
+            if group in self.plot_behind:
+                z = 0
+            else:
+                z = en + 1
+            xu = self.adata[group_df.index].obsm[self._use_key]
+            self.axes[0].scatter(
+                xu[:, 0],
+                xu[:, 1],
+                label=group,
+                color=self.CMAP[group],
+                alpha=self._alpha,
+                zorder = z,
+            )
+
+    def _format(self):
+        for ax in self.axes:
+            ax.set_xlabel(self._labels[0])
+            ax.set_ylabel(self._labels[1])
+        self.axes[0].legend(loc=(1, 0.25), edgecolor="w")
+        self.axes[0].set_title(self._title)
+
+    def __call__(
+        self,
+        groupby=None,
+        plot_behind = ["Undifferentiated"],
+    ):
+        self.__update__(locals())
+
+        self.__config__()
+        self._plot_grouped()
+        self._format()

torch_adata/_utils/__init__.py

@@ -0,0 +1,4 @@
+
+from ._abc_parse import ABCParse
+from ._info_message import InfoMessage
+from ._function_inspector import FunctionInspector

torch_adata/_utils/_abc_parse.py

@@ -0,0 +1,73 @@
+from abc import ABC
+from typing import Dict, List, Any
+NoneType = type(None)
+
+class ABCParse(ABC):
+    _BUILT = False
+
+    def __init__(self, *args, **kwargs):
+        """
+        we avoid defining things in __init__ because this subsequently
+        mandates the use of `super().__init__()`
+        """
+        pass
+
+    def __build__(self) -> None:
+
+        self._PARAMS = {}
+        self._IGNORE = ["self", "__class__"]
+        self._stored_private = []
+        self._stored_public = []
+
+        self._BUILT = True
+
+    def __set__(
+        self, key: str, val: Any, public: List = [], private: List = []
+    ) -> None:
+
+        self._PARAMS[key] = val
+        if (key in private) and (not key in public):
+            self._stored_private.append(key)
+            key = f"_{key}"
+        else:
+            self._stored_public.append(key)
+        setattr(self, key, val)
+
+    def __set_existing__(self, key: str, val: Any) -> None:
+
+        self._PARAMS[key] = val
+
+        if key in self._stored_private:
+            key = f"_{key}"
+        setattr(self, key, val)
+
+    @property
+    def _STORED(self) -> List:
+        return self._stored_private + self._stored_public
+
+    def __parse__(
+        self, kwargs: Dict, public: List = [], private: List = [], ignore: List = []
+    ):
+
+        if not self._BUILT:
+            self.__build__()
+
+        self._IGNORE += ignore
+
+        if len(public) > 0:
+            private = list(kwargs.keys())
+
+        for key, val in kwargs.items():
+            if not key in self._IGNORE:
+                self.__set__(key, val, public, private)
+
+    def __update__(self, kwargs: dict, public: List = [], private: List = []) -> None:
+
+        if not self._BUILT:
+            self.__build__()
+
+        for key, val in kwargs.items():
+            if not isinstance(val, NoneType) and key in self._STORED:
+                self.__set_existing__(key, val)
+            elif not isinstance(val, NoneType) and not key in self._IGNORE:
+                self.__set__(key, val, public, private)

torch_adata/_utils/_function_inspector.py

@@ -0,0 +1,53 @@
+
+
+import inspect
+
+
+class FunctionInspector:
+    _FUNC_KWARGS = {}
+
+    def __init__(self, func, ignore=[]):
+        """
+        Parameters:
+        -----------
+        func
+            type: Any
+
+        ignore
+            type: list
+
+        Returns:
+        --------
+        None
+        """
+        self._func = func
+        self._ignore = ignore
+
+    def _extract_func_params(self, func):
+        return list(inspect.signature(func).parameters.keys())
+
+    @property
+    def FUNC_PARAMS(self):
+        return self._extract_func_params(self._func)
+
+    def forward(self, key, val):
+        if (key in self.FUNC_PARAMS) and (not key in self._ignore):
+            self._FUNC_KWARGS[key] = val
+
+    def __call__(self, kwargs) -> dict:
+        """
+        Parameters:
+        -----------
+        kwargs
+            type: dict
+
+        Returns:
+        --------
+        func_kwargs
+            type: dict
+        """
+
+        for key, val in kwargs.items():
+            self.forward(key, val)
+
+        return self._FUNC_KWARGS