diff --git a/src/db_utils/__main__.py b/src/db_utils/__main__.py
index c6679832c5094d22f02a79a4ebed164354dd3346..55599b1901a34182150345e6a85f39ededde7620 100755
--- a/src/db_utils/__main__.py
+++ b/src/db_utils/__main__.py
@@ -10,8 +10,7 @@ from .config import Config
from ..logging_conf import logging_def
from .db_creation import main_create_db
from .fill_exon_n_gene_table import main_fill_exon_n_gene
-# from .create_freq_table import create_freq_table
-# from .populate_database import populate
+from .fill_splicing_lore_tables import fill_splicing_lore_data
import logging
@@ -24,21 +23,8 @@ def launcher(logging_level: str = "INFO"):
main_create_db('DISABLE')
logging.info('Filling cin_gene and cin_exon')
main_fill_exon_n_gene('DISABLE')
- # logging.info(f"Creation of {Config.ctrl_exon_file} file")
- # get_ctrl(Config.exon_intern)
- # logging.info(f"Creation of {Config.exon_file} file")
- # get_exon_table(Config.ctrl_exon_file, Config.gene_file, logging_level)
- # logging.info(f"Creation of {Config.frequency_file} file")
- # create_freq_table(Config.bed_orf, Config.bed_exon, Config.ctrl_exon_file,
- # logging_level)
- #
- #Â mpopulate = populate.__wrapped__
- # logging.info(f"Filling {Config.tables[0]} table")
- # mpopulate(Config.tables[0], Config.gene_file, "y", logging_level)
- # logging.info(f"Filling {Config.tables[1]} table")
- # mpopulate(Config.tables[1], Config.exon_file, "y", logging_level)
- # logging.info(f"Filling {Config.tables[2]} table")
- # mpopulate(Config.tables[2], Config.frequency_file, "y", logging_level)
+ logging.info('Filling splicing lore tables')
+ fill_splicing_lore_data('DISABLE')
-launcher(logging_level = "DEBUG")
\ No newline at end of file
+launcher(logging_level="DEBUG")
diff --git a/src/db_utils/config.py b/src/db_utils/config.py
index 8f9d60ba71004f18633749eb5d45fce0de6e70ff..6155a2f44a8e960db1e23d9c4b26b064c6f54d68 100755
--- a/src/db_utils/config.py
+++ b/src/db_utils/config.py
@@ -13,8 +13,11 @@ class Config:
"""
A class containing every parameters used in the submodule db_utils
"""
- db_file = Path(__file__).parents[2] / "results" / 'chia_pet_database.db'
- tables = ["cin_gene", "cin_exon", "cin_frequency", "cin_interaction"]
- bed_exon = Path(__file__).parents[2] / 'data' / 'bed' / 'exon.bed'
- bed_gene = Path(__file__).parents[2] / 'data' / 'bed' / 'gene.bed'
-
+ data = Path(__file__).parents[2] / 'data'
+ results = Path(__file__).parents[2] / "results"
+ db_file = results / 'chia_pet_database.db'
+ bed_exon = data / 'bed' / 'exon.bed'
+ bed_gene = data / 'bed' / 'gene.bed'
+ ase_event_file = data / 'splicing_lore_data' / 'ase_event.txt'
+ splicing_projects = data / 'splicing_lore_data' / \
+ 'splicing_lore_projects.txt'
diff --git a/src/db_utils/fill_splicing_lore_tables.py b/src/db_utils/fill_splicing_lore_tables.py
new file mode 100644
index 0000000000000000000000000000000000000000..63cadbb06d90a534c7f390504ee95b5688b3c121
--- /dev/null
+++ b/src/db_utils/fill_splicing_lore_tables.py
@@ -0,0 +1,63 @@
+#!/usr/bin/env python3
+
+# -*- coding: UTF-8 -*-
+
+"""
+Description: The goal of this script is to fill the tables \
+cin_ase_event and cin_project_splicing_lore
+"""
+
+from .config import Config
+import pandas as pd
+import logging
+from ..logging_conf import logging_def
+from .populate_database import populate_df
+
+
+def load_splicing_projects() -> pd.DataFrame:
+ """
+ Load the projects of splicing lore database.
+
+ :return: The dataframe of the splicing lore projects
+ """
+ df = pd.read_csv(Config.splicing_projects, sep="\t")
+ logging.debug(df.head())
+ return df
+
+
+def load_ase_event() -> pd.DataFrame:
+ """
+ Load the alternative splicing events of detected in the splicing lore \
+ database.
+
+ :return: The dataframe of splicing lore alternative splicing events
+ """
+ logging.debug('loading cin_project_splicing_lore')
+ df = pd.read_csv(Config.ase_event_file, sep="\t")
+ df.drop(["gene_symbol", "chromosome", "start", "stop", "exons_flanquants"],
+ inplace=True, axis=1)
+ cols = list(df.columns)
+ if cols.index('exon_skipped') != -1:
+ cols[cols.index('exon_skipped')] = 'pos'
+ df.columns = cols
+ df['exon_id'] = df['gene_id'].astype(str) + '_' + df['pos'].astype(str)
+ logging.debug(df.head())
+ return df
+
+
+def fill_splicing_lore_data(logging_level: str = 'DISABLE') -> None:
+ """
+ Fill the tables cin_ase_event and cin_project_splicing_lore
+ """
+ logging_def(Config.results, __file__, logging_level)
+
+ sf_projects = load_splicing_projects()
+ ase_events = load_ase_event()
+ logging.debug('Filling cin_project_splicing_lore')
+ populate_df(table='cin_project_splicing_lore', df=sf_projects, clean='y')
+ logging.debug('Filling cin_ase_event')
+ populate_df(table='cin_ase_event', df=ase_events, clean='y')
+
+
+if __name__ == "__main__":
+ fill_splicing_lore_data('DEBUG')
diff --git a/src/db_utils/populate_database.py b/src/db_utils/populate_database.py
index 56ff2750217e745a2cf2131d62fe80a65d563950..4fcf80ff8e7016ac9c37428f4d207ec44effd636 100755
--- a/src/db_utils/populate_database.py
+++ b/src/db_utils/populate_database.py
@@ -113,16 +113,31 @@ def insert_data(table: str, content: List[Tuple], cnx: sqlite3.Connection
cursor.close()
-def check_table_name(table : str):
+def get_table_names(cnx: sqlite3.Connection) -> List[str]:
+ """
+ Get the list of available table names.
+
+ :param cnx: The connection to ChIA-PET database.
+
+ :return: The list of availbale tables
+ """
+ c = cnx.cursor()
+ c.execute("""SELECT name FROM sqlite_master WHERE type = 'table';""")
+ res = c.fetchall()
+ res = [r[0] for r in res]
+ return res
+
+
+def check_table_name(cnx: sqlite3.Connection, table: str):
"""
Check if we can use the table name `table`.
+ :param cnx: Connection to ChIA-PET database
:param table: The name of the table to fille
:return: The same name with the prefix gin if it wasn't here.
"""
- if "cin" not in table:
- table = f"cin_{table.lower()}"
- if table not in Config.tables:
+ tables = get_table_names(cnx)
+ if table not in tables:
msg = f"The name {table} is not available." \
f" If the table exist in the database, " \
f"change the config file to add the table name " \
@@ -175,8 +190,8 @@ def populate_df(table: str, df: pd.DataFrame, clean: str):
tab.
:param clean: y to remove the data in the table, n else.
"""
- table = check_table_name(table)
cnx = sqlite3.connect(Config.db_file)
+ table = check_table_name(cnx, table)
check_content_clean_and_insert(table, df, cnx, clean)
@@ -195,8 +210,8 @@ def populate(table: str, file: str, clean: str, logging_level: str =
"""
logging_def(Config.db_file.parent, __file__, logging_level)
mfile = Path(file)
- table = check_table_name(table)
cnx = sqlite3.connect(Config.db_file)
+ table = check_table_name(cnx, table)
check_content_clean_and_insert(table, mfile, cnx, clean)