From 5975e7c33a9c9743b2604896b41e63df07495e05 Mon Sep 17 00:00:00 2001
From: Fontrodona Nicolas <nicolas.fontrodona@ens-lyon.fr>
Date: Wed, 4 Mar 2020 10:07:18 +0100
Subject: [PATCH] src/db_utils/fill_splicing_lore_tables.py,
src/db_utils/__main__.py: creation of fill_splicing_lore_tables.py that fills
cin_ase_event and cin_project_splicing_lore tables
---
src/db_utils/__main__.py | 22 ++------
src/db_utils/fill_splicing_lore_tables.py | 63 +++++++++++++++++++++++
2 files changed, 67 insertions(+), 18 deletions(-)
create mode 100644 src/db_utils/fill_splicing_lore_tables.py
diff --git a/src/db_utils/__main__.py b/src/db_utils/__main__.py
index c6679832..55599b19 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/fill_splicing_lore_tables.py b/src/db_utils/fill_splicing_lore_tables.py
new file mode 100644
index 00000000..63cadbb0
--- /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')
--
GitLab