From 3d00a27da2dc1ae9cb8cb8169cdf8aa6ccfd87fb Mon Sep 17 00:00:00 2001
From: Remi PLANEL <rplanel@pasteur.fr>
Date: Wed, 5 Jun 2024 11:30:19 +0200
Subject: [PATCH] Eleos special case
---
.../df-wiki-cli/df_wiki_cli/content/main.py | 114 ++++++++++++------
1 file changed, 76 insertions(+), 38 deletions(-)
diff --git a/packages/df-wiki-cli/df_wiki_cli/content/main.py b/packages/df-wiki-cli/df_wiki_cli/content/main.py
index 8bef84e4..1882aece 100644
--- a/packages/df-wiki-cli/df_wiki_cli/content/main.py
+++ b/packages/df-wiki-cli/df_wiki_cli/content/main.py
@@ -304,44 +304,82 @@ def system_operon_structure(
in_exchangeables = False
current_gene = {}
exchangeables = []
- with open(def_path["path"]) as file:
- for event, elem in ET.iterparse(file, events=("start", "end")):
- if event == "start":
- if (
- elem.tag == "gene"
- and not in_exchangeables
- and elem.attrib["presence"] != "forbidden"
- ):
- current_gene = {
- "system": system,
- "subsystem": subsystem,
- "gene": elem.attrib["name"],
- "version": def_path["version"],
- "exchangeables": None,
- }
-
- # get hmm definition to extract the size
- base_path_list = def_path["path"].parts[0:-4]
- profile_path = (
- Path(*base_path_list)
- / f"profiles/{current_gene['gene']}.hmm"
- )
- with open(profile_path) as profile_file:
- for line in profile_file:
- if line.startswith("LENG"):
- size = int(re.split(r"\s+", line)[1])
- current_gene["size"] = size
- system_genes.append(current_gene)
- if elem.tag == "gene" and in_exchangeables:
- exchangeables.append(elem.attrib["name"])
- if elem.tag == "exchangeables":
- in_exchangeables = True
- exchangeables = []
- elif event == "end":
- if elem.tag == "exchangeables":
- in_exchangeables = False
- current_gene["exchangeables"] = ",".join(exchangeables)
- exchangeables = []
+
+ # <model inter_gene_max_space="5" min_mandatory_genes_required="2" min_genes_required="2" vers="2.0">
+
+ # <gene name="Eleos__LeoA" presence="mandatory">
+ # <exchangeables>
+ # <gene name="Eleos__LeoA2"/>
+ # </exchangeables>
+ # </gene>
+
+ # <gene name="Eleos__LeoBC" presence="mandatory"/>
+
+ # <gene name="Eleos__LeoB" presence="mandatory"/>
+
+ # <gene name="Eleos__LeoC" presence="mandatory"/>
+
+ # </model>
+ if system == 'Eleos':
+ eleos_struct = [
+ {
+ "system": system,
+ "subsystem": "Eleos",
+ "gene": "Eleos__LeoA",
+ "version": "1.2.4",
+ "exchangeables": "Eleos__LeoA2",
+ "size": 559,
+ },
+ {
+ "system": system,
+ "subsystem": "Eleos",
+ "gene": "Eleos__LeoBC",
+ "version": "1.2.4",
+ "exchangeables": None,
+ "size": 601,
+ },
+
+ ]
+ system_genes += eleos_struct
+ else:
+ with open(def_path["path"]) as file:
+ for event, elem in ET.iterparse(file, events=("start", "end")):
+ if event == "start":
+ if (
+ elem.tag == "gene"
+ and not in_exchangeables
+ and elem.attrib["presence"] != "forbidden"
+ ):
+ current_gene = {
+ "system": system,
+ "subsystem": subsystem,
+ "gene": elem.attrib["name"],
+ "version": def_path["version"],
+ "exchangeables": None,
+ }
+
+ # get hmm definition to extract the size
+ base_path_list = def_path["path"].parts[0:-4]
+ profile_path = (
+ Path(*base_path_list)
+ / f"profiles/{current_gene['gene']}.hmm"
+ )
+ with open(profile_path) as profile_file:
+ for line in profile_file:
+ if line.startswith("LENG"):
+ size = int(re.split(r"\s+", line)[1])
+ current_gene["size"] = size
+ system_genes.append(current_gene)
+ if elem.tag == "gene" and in_exchangeables:
+ exchangeables.append(elem.attrib["name"])
+ if elem.tag == "exchangeables":
+ in_exchangeables = True
+ exchangeables = []
+ elif event == "end":
+ if elem.tag == "exchangeables":
+ in_exchangeables = False
+ current_gene["exchangeables"] = ",".join(exchangeables)
+ exchangeables = []
with open(output, "w") as f:
fieldnames = [
--
GitLab