init source

public/jsonpathway.py

@@ -0,0 +1,83 @@
+import xml.etree.ElementTree as ET
+import json
+
+def parse_kegg_xml_with_group(xml_file):
+    tree = ET.parse(xml_file)
+    root = tree.getroot()
+
+    pathway_info = {
+        "name": root.attrib.get("name"),
+        "org": root.attrib.get("org"),
+        "number": root.attrib.get("number"),
+        "title": root.attrib.get("title"),
+        "image": root.attrib.get("image"),
+        "link": root.attrib.get("link"),
+        "entries": [],
+        "relations": [],
+        "reactions": []
+    }
+
+    for entry in root.findall("entry"):
+        entry_id = int(entry.attrib.get("id"))
+        entry_group = str((entry_id // 10000) * 10000) if entry_id >= 10000 else None
+        graphics = entry.find("graphics")
+        entry_data = {
+            "id": entry_id,
+            "name": entry.attrib.get("name"),
+            "type": entry.attrib.get("type"),
+            "link": entry.attrib.get("link"),
+            "reaction": entry.attrib.get("reaction"),
+            "group": entry_group,
+            "graphics": {
+                "name": graphics.attrib.get("name"),
+                "fgcolor": graphics.attrib.get("fgcolor"),
+                "bgcolor": graphics.attrib.get("bgcolor"),
+                "type": graphics.attrib.get("type"),
+                "x": int(graphics.attrib.get("x")),
+                "y": int(graphics.attrib.get("y")),
+                "width": int(graphics.attrib.get("width")),
+                "height": int(graphics.attrib.get("height")),
+            }
+        }
+        pathway_info["entries"].append(entry_data)
+
+    for relation in root.findall("relation"):
+        rel = {
+            "entry1": int(relation.attrib.get("entry1")),
+            "entry2": int(relation.attrib.get("entry2")),
+            "type": relation.attrib.get("type"),
+            "subtypes": []
+        }
+        for subtype in relation.findall("subtype"):
+            rel["subtypes"].append({
+                "name": subtype.attrib.get("name"),
+                "value": subtype.attrib.get("value")
+            })
+        pathway_info["relations"].append(rel)
+
+    for reaction in root.findall("reaction"):
+        reac = {
+            "id": int(reaction.attrib.get("id")),
+            "name": reaction.attrib.get("name"),
+            "type": reaction.attrib.get("type"),
+            "substrates": [],
+            "products": []
+        }
+        for substrate in reaction.findall("substrate"):
+            reac["substrates"].append({
+                "id": int(substrate.attrib.get("id")),
+                "name": substrate.attrib.get("name")
+            })
+        for product in reaction.findall("product"):
+            reac["products"].append({
+                "id": int(product.attrib.get("id")),
+                "name": product.attrib.get("name")
+            })
+        pathway_info["reactions"].append(reac)
+
+    return pathway_info
+
+# 사용 예:
+result = parse_kegg_xml_with_group("expanded_pathway43200.xml")
+with open("group43200.json", "w", encoding="utf-8") as f:
+    json.dump(result, f, ensure_ascii=False, indent=2)

public/makepathway.py

@@ -0,0 +1,52 @@
+import xml.etree.ElementTree as ET
+import copy
+
+# 원본 XML 파일 경로
+input_file = 'pon00061.xml'
+output_file = 'expanded_pathway43200.xml'
+
+# 파싱
+tree = ET.parse(input_file)
+root = tree.getroot()
+
+# entry 태그만 추출
+entries = [e for e in root.findall('entry')]
+
+# 원본 entry 수 확인
+print(f'원본 entry 개수: {len(entries)}')
+
+# 복제 단계 수: 10000~90000 까지 (1~9단계)
+MULTIPLIERS = range(1, 120)
+
+# 새로운 entry 리스트
+new_entries = []
+
+for mult in MULTIPLIERS:
+    id_offset = mult * 10000
+    xy_offset = mult * 2000
+
+    for entry in entries:
+        new_entry = copy.deepcopy(entry)
+        
+        # id 업데이트
+        old_id = int(entry.attrib['id'])
+        new_entry.attrib['id'] = str(old_id + id_offset)
+
+        # graphics 내부 x, y 업데이트
+        graphics = new_entry.find('graphics')
+        if graphics is not None:
+            old_x = int(graphics.attrib['x'])
+            old_y = int(graphics.attrib['y'])
+
+            graphics.attrib['x'] = str(old_x + xy_offset)
+            graphics.attrib['y'] = str(old_y + xy_offset)
+
+        new_entries.append(new_entry)
+
+# 원본 root에 추가
+for e in new_entries:
+    root.append(e)
+
+# 출력 저장
+tree.write(output_file, encoding='utf-8', xml_declaration=True)
+print(f'새로운 XML이 {output_file}에 저장되었습니다.')

public/pon00062.xml

@@ -0,0 +1,39 @@
+<?xml version="1.0"?>
+<!DOCTYPE pathway SYSTEM "https://www.kegg.jp/kegg/xml/KGML_v0.7.2_.dtd">
+<!-- Creation date: Jul 1, 2024 17:49:43 +0900 (GMT+9) -->
+<pathway name="path:pon00061" org="pon" number="00061"
+         title="Fatty acid biosynthesis"
+         image="https://www.kegg.jp/kegg/pathway/pon/pon00061.png"
+         link="https://www.kegg.jp/kegg-bin/show_pathway?pon00061">
+    <entry id="360" name="pon:100462086" type="gene" reaction="rn:R02767"
+        link="https://www.kegg.jp/dbget-bin/www_bget?pon:100462086">
+        <graphics name="CBR4" fgcolor="#000000" bgcolor="#BFFFBF"
+             type="rectangle" x="981" y="1260" width="46" height="17"/>
+    </entry>
+    <entry id="361" name="pon:100461538" type="gene"
+        link="https://www.kegg.jp/dbget-bin/www_bget?pon:100461538">
+        <graphics name="HSD17B8" fgcolor="#000000" bgcolor="#BFFFBF"
+             type="rectangle" x="981" y="1277" width="46" height="17"/>
+    </entry>
+    <entry id="367" name="cpd:C05744 cpd:C00685" type="compound"
+        link="https://www.kegg.jp/dbget-bin/www_bget?C05744+C00685">
+        <graphics name="C05744..." fgcolor="#000000" bgcolor="#FFFFFF"
+             type="circle" x="956" y="1231" width="8" height="8"/>
+    </entry>
+    <entry id="426" name="pon:100438452" type="gene" reaction="rn:R04355"
+        link="https://www.kegg.jp/dbget-bin/www_bget?pon:100438452">
+        <graphics name="OXSM" fgcolor="#000000" bgcolor="#BFFFBF"
+             type="rectangle" x="855" y="1193" width="46" height="17"/>
+    </entry>
+    <relation entry1="426" entry2="360" type="ECrel">
+        <subtype name="compound" value="367"/>
+    </relation>
+    <relation entry1="426" entry2="361" type="ECrel">
+        <subtype name="compound" value="367"/>
+    </relation>
+    <reaction id="426" name="rn:R04355" type="irreversible">
+        <substrate id="425" name="cpd:C03939"/>
+        <substrate id="366" name="cpd:C01209"/>
+        <product id="367" name="cpd:C05744"/>
+    </reaction>
+</pathway>

public/robots.txt

@@ -0,0 +1,2 @@
+User-Agent: *
+Disallow: