Merge branch 'main' into tests

bedapub · Dec 8, 2024 · 1b3cdcf · 1b3cdcf
2 parents 1e4f231 + 650bff8
commit 1b3cdcf
Show file tree

Hide file tree

Showing 6 changed files with 44 additions and 51 deletions.
diff --git a/docs/chapters/features/usage.rst b/docs/chapters/features/usage.rst
@@ -1,46 +1,47 @@
-Import package and initialise Kinex 
+Import Package and Initialize Kinex 
 ===================================
 
-1. Import kinex
+1. **Import Kinex**
 
 .. code:: python
 
-	from kinex import Kinex
+    from kinex import Kinex
 
-2. Read the scoring matrix
+2. **Create a Kinex Object**
+
+- With Predefined Matrices:
+
+It will look in the resources to find the matrices. If it doesn't find them, it will download them and save them for future use.
 
 .. code:: python
 
-    scoring_matrix_ser_thr = pd.read_csv("https://zenodo.org/records/13964893/files/scoring_matrix_ser_thr_82k_sorted.csv.gz?download=1", compression="gzip")
-    scoring_matrix_tyr = pd.read_csv("https://zenodo.org/records/13964893/files/scoring_matrix_tyr_7k_sorted.csv.gz?download=1", compression="gzip")
-    scoring_matrix_ser_thr
+    kinex = Kinex()
+
+- With Your Custom Matrices:
 
 .. code:: python
 
-                AAK1    ACVR2A  ...      YSK4       ZAK
-    0     -11.147481 -6.325340  ... -6.723077 -7.402360
-    1     -10.421859 -6.178601  ... -6.343452 -7.373478
-    ...          ...       ...  ...
-    82753   8.074270  7.289390  ...  4.525527  4.837377
-    82754   8.623180  7.871226  ...  4.869195  5.062391
+    scoring_matrix_ser_thr = pd.read_csv("path/to/scoring_matrix_ser_thr.csv")
+    scoring_matrix_tyr = pd.read_csv("path/to/scoring_matrix_tyr.csv")
 
-    [82755 rows x 303 columns]
+    kinex = Kinex(scoring_matrix_ser_thr, scoring_matrix_tyr)
 
 .. note::
 
-    You can optionally save the scoring matrix locally for faster use in the future.
+    The matrix looks like this:
 
     .. code:: python
 
-        scoring_matrix_ser_thr.to_csv("scoring_matrix_ser_thr.csv")
-        scoring_matrix_tyr.to_csv("scoring_matrix_tyr.csv")
+                      AAK1    ACVR2A  ...      YSK4       ZAK
+        0     -11.147481 -6.325340  ... -6.723077 -7.402360
+        1     -10.421859 -6.178601  ... -6.343452 -7.373478
+        ...          ...       ...  ...
+        82753   8.074270  7.289390  ...  4.525527  4.837377
+        82754   8.623180  7.871226  ...  4.869195  5.062391
 
-    Or just download using the links: 
-    `https://zenodo.org/records/13964893/files/scoring_matrix_ser_thr_82k_sorted.csv.gz?download=1 <https://zenodo.org/records/13964893/files/scoring_matrix_ser_thr_82k_sorted.csv.gz?download=1>`_
-    `https://zenodo.org/records/13964893/files/scoring_matrix_tyr_7k_sorted.csv.gz?download=1 <https://zenodo.org/records/13964893/files/scoring_matrix_tyr_7k_sorted.csv.gz?download=1>`_
-
-3.  Create a kinex object
+.. note::
 
-.. code:: python
+    Predefined matrices can be found here:
 
-    kinex = Kinex(scoring_matrix_ser_thr, scoring_matrix_tyr)
+    - `Scoring Matrix for Serine/Threonine <https://zenodo.org/records/13964893/files/scoring_matrix_ser_thr_82k_sorted.csv.gz?download=1>`_
+    - `Scoring Matrix for Tyrosine <https://zenodo.org/records/13964893/files/scoring_matrix_tyr_7k_sorted.csv.gz?download=1>`_
diff --git a/pyproject.toml b/pyproject.toml
@@ -20,7 +20,7 @@ dependencies = [
     "scikit-learn",
     "umap-learn",
     "importlib-resources",
-    "tomli",
+    "requests",
 ]
 
 [project.optional-dependencies]
@@ -29,10 +29,6 @@ dev = [
     "furo"
 ]
 
-[project.urls]
-scoring_matrix_ser_thr = "https://zenodo.org/records/13964893/files/scoring_matrix_ser_thr_82k_sorted.csv.gz?download=1"
-scoring_matrix_tyr = "https://zenodo.org/records/13964893/files/scoring_matrix_tyr_7k_sorted.csv.gz?download=1"
-
 
 [tool.setuptools]
 package-dir = { "" = "src" }

diff --git a/requirements.txt b/requirements.txt
@@ -6,5 +6,4 @@ statsmodels
 plotly
 scikit-learn
 umap-learn
-importlib-resources
-tomli
+importlib-resources
diff --git a/src/kinex/kinex.py b/src/kinex/kinex.py
@@ -1,19 +1,13 @@
 import bisect
 from collections import namedtuple
 from functools import reduce
-import tomli
 
 import numpy as np
 import pandas as pd
 
 from kinex.enrichment import Enrichment
 from kinex.functions import download_file_to_resource
-from kinex.resources import (
-    get_pssm_ser_thr,
-    get_pssm_tyr,
-    get_scoring_matrix_ser_thr,
-    get_scoring_matrix_tyr,
-)
+from kinex.resources import get_pssm_ser_thr, get_pssm_tyr, get_scoring_matrix_ser_thr, get_scoring_matrix_tyr, get_configuration_file
 from kinex.score import Score
 from kinex.enrichment import Enrichment
 from kinex.sequence import get_sequence_object, SequenceType
@@ -23,8 +17,7 @@
 
 
 # Load the pyproject.toml file
-with open("pyproject.toml", "rb") as f:
-    config = tomli.load(f)
+config = get_configuration_file()
 
 
 class Kinex:
@@ -99,21 +92,15 @@ def __init__(
             scoring_matrix_ser_thr = get_scoring_matrix_ser_thr()
             # Matrix is not provided and not found in the resources, download the default matrix
             if scoring_matrix_ser_thr is None:
-                scoring_matrix_ser_thr_url = config["project"]["urls"][
-                    "scoring_matrix_ser_thr"
-                ]
-                download_file_to_resource(
-                    scoring_matrix_ser_thr_url, "default_scoring_matrix_ser_thr.csv.gz"
-                )
+                scoring_matrix_ser_thr_url = config["urls"]["scoring_matrix_ser_thr"]
+                download_file_to_resource(scoring_matrix_ser_thr_url, 'default_scoring_matrix_ser_thr.csv.gz')
                 scoring_matrix_ser_thr = get_scoring_matrix_ser_thr()
 
         if scoring_matrix_tyr is None:
             scoring_matrix_tyr = get_scoring_matrix_tyr()
             if scoring_matrix_tyr is None:
-                scoring_matrix_tyr_url = config["project"]["urls"]["scoring_matrix_tyr"]
-                download_file_to_resource(
-                    scoring_matrix_tyr_url, "default_scoring_matrix_tyr.csv.gz"
-                )
+                scoring_matrix_tyr_url = config["urls"]["scoring_matrix_tyr"]
+                download_file_to_resource(scoring_matrix_tyr_url, 'default_scoring_matrix_tyr.csv.gz')
                 scoring_matrix_tyr = get_scoring_matrix_tyr()
 
         self.pssm_ser_thr = pssm_ser_thr

diff --git a/src/kinex/resources/__init__.py b/src/kinex/resources/__init__.py
@@ -49,4 +49,8 @@ def get_scoring_matrix_tyr() -> pd.DataFrame:
         with resources.files("kinex.resources").joinpath("default_scoring_matrix_tyr.csv.gz").open('rb') as file_path:
             return pd.read_csv(file_path, compression='gzip')
     except FileNotFoundError:
-        return None
+        return None
+
+def get_configuration_file() -> dict:
+    with resources.files("kinex.resources").joinpath("config.json").open() as json_file:
+        return json.load(json_file)
diff --git a/src/kinex/resources/config.json b/src/kinex/resources/config.json
@@ -0,0 +1,6 @@
+{
+  "urls": {
+    "scoring_matrix_ser_thr": "https://zenodo.org/records/13964893/files/scoring_matrix_ser_thr_82k_sorted.csv.gz?download=1",
+    "scoring_matrix_tyr": "https://zenodo.org/records/13964893/files/scoring_matrix_tyr_7k_sorted.csv.gz?download=1"
+  }
+}