feat: added base constants

Author: PauliusPreiksaCode

data_collection/constants.py

@@ -1,15 +1,19 @@
 """Module for constants used in data collection."""
 
 # files
-LOVD_URL = "https://databases.lovd.nl/shared/genes/EYS"
-LOVD_FILE_URL = "https://databases.lovd.nl/shared/download/all/gene/EYS"
+LOVD_URL = "https://databases.lovd.nl/shared/genes"
+LOVD_URL_EYS = "https://databases.lovd.nl/shared/genes/EYS"
+LOVD_FILE_URL = "https://databases.lovd.nl/shared/download/all/gene"
+LOVD_FILE_URL_EYS = "https://databases.lovd.nl/shared/download/all/gene/EYS"
 
-GNOMAD_URL = "https://gnomad.broadinstitute.org/gene/ENSG00000188107?dataset=gnomad_r4"
-GNOMAD_FILE_URL = ("https://drive.usercontent.google.com/u/0/uc?id=1crkDCVcC0PSnv0JPGj3FpemBs28"
+GNOMAD_URL = "https://gnomad.broadinstitute.org/gene"
+GNOMAD_URL_EYS = "https://gnomad.broadinstitute.org/gene/ENSG00000188107?dataset=gnomad_r4"
+GNOMAD_FILE_URL_EYS = ("https://drive.usercontent.google.com/u/0/uc?id=1crkDCVcC0PSnv0JPGj3FpemBs28"
                    "-T_3y&export=download")
 
-CLINVAR_URL = "https://www.ncbi.nlm.nih.gov/clinvar/?term=eys%5Bgene%5D&redir=gene"
-CLINVAR_FILE_URL = ("https://drive.usercontent.google.com/u/0/uc?id=1RK5XBK3k5h0K6f-qfwJSQj7tlF"
+CLINVAR_URL = "https://www.ncbi.nlm.nih.gov/clinvar"
+CLINVAR_URL_EYS = "https://www.ncbi.nlm.nih.gov/clinvar/?term=eys%5Bgene%5D&redir=gene"
+CLINVAR_FILE_URL_EYS = ("https://drive.usercontent.google.com/u/0/uc?id=1RK5XBK3k5h0K6f-qfwJSQj7tlF"
                     "-H2U6u&export=download")
 
 # paths

data_collection/pipeline.py

@@ -2,9 +2,9 @@
 import pandas as pd
 
 from tools import get_file_from_url, from_lovd_to_pandas, from_clinvar_name_to_dna
-from constants import (LOVD_FILE_URL,
-                       GNOMAD_FILE_URL,
-                       CLINVAR_FILE_URL,
+from constants import (LOVD_FILE_URL_EYS,
+                       GNOMAD_FILE_URL_EYS,
+                       CLINVAR_FILE_URL_EYS,
                        DATA_PATH,
                        LOVD_PATH,
                        GNOMAD_PATH,
@@ -47,9 +47,9 @@ def calculate_max_frequency(row):
 
 # MAIN
 # Download all data
-get_file_from_url(LOVD_FILE_URL, LOVD_PATH + "/lovd_data.txt", override=True)
-get_file_from_url(GNOMAD_FILE_URL, GNOMAD_PATH + "/gnomad_data.csv", override=True)
-get_file_from_url(CLINVAR_FILE_URL, CLINVAR_PATH + "/clinvar_data.txt", override=True)
+get_file_from_url(LOVD_FILE_URL_EYS, LOVD_PATH + "/lovd_data.txt", override=True)
+get_file_from_url(GNOMAD_FILE_URL_EYS, GNOMAD_PATH + "/gnomad_data.csv", override=True)
+get_file_from_url(CLINVAR_FILE_URL_EYS, CLINVAR_PATH + "/clinvar_data.txt", override=True)
 
 # Read and convert data
 lovd_data = from_lovd_to_pandas(LOVD_PATH + "/lovd_data.txt")