Adding GO annotation function based on GO id

Juse_toolkit.py

@@ -991,11 +991,11 @@ def setupUi(self, Dialog):
         self.enrich_xlab.addItem("")
         self.horizontalLayout_38.addWidget(self.enrich_xlab)
         self.gtl_dl = QtWidgets.QPushButton(self.groupBox_31)
-        self.gtl_dl.setEnabled(False)
+        self.gtl_dl.setEnabled(True)
         self.gtl_dl.setObjectName("gtl_dl")
         self.horizontalLayout_38.addWidget(self.gtl_dl)
         self.gef_c = QtWidgets.QPushButton(self.groupBox_31)
-        self.gef_c.setEnabled(False)
+        self.gef_c.setEnabled(True)
         self.gef_c.setObjectName("gef_c")
         self.horizontalLayout_38.addWidget(self.gef_c)
         self.groupBox_32 = QtWidgets.QGroupBox(self.tab_8)

Juse_toolkit.ui

@@ -3044,7 +3044,7 @@ p, li { white-space: pre-wrap; }
          <item>
           <widget class="QPushButton" name="gtl_dl">
            <property name="enabled">
-            <bool>false</bool>
+            <bool>true</bool>
            </property>
            <property name="text">
             <string>下载 GO term list 文件</string>
@@ -3054,7 +3054,7 @@ p, li { white-space: pre-wrap; }
          <item>
           <widget class="QPushButton" name="gef_c">
            <property name="enabled">
-            <bool>false</bool>
+            <bool>true</bool>
            </property>
            <property name="text">
             <string>对 GO 富集文件进行分类</string>

README.md

@@ -17,3 +17,20 @@
 教程可见：[Juse's blog](https://jusetiz.github.io/)
 
 若存在报错 Bug 可以联系 Juse 修复。
+
+目前已经实现的功能：
+
+- 提取最长转录本。*（已实现）*
+- 根据 id 提取序列。*（已实现）*
+- 对序列的 id 进行各种处理。*（已实现）*
+- 串联序列并得到分区信息。*（已实现）*
+- 批量进行序列格式转换。*（已实现）*
+- 批量提取 Orthofinder 的 orthogroup 对应的 CDS 序列。*（已实现）*
+- 批量进行序列的物种数和长度过滤。*（已实现）*
+- 火山图绘制。*（已实现）*
+- 气泡图绘制。*（已实现）*
+
+目前已有的小功能：
+
+- 学习计时器。
+- 批量改后缀。

plot.py

@@ -148,4 +148,32 @@ def plot_GOem(file_path, num = 20, xlab = 'GeneRatio', ylab = 'Description'):
     plt.ylabel('GO Description', fontsize=12)
 
     plt.tight_layout()
-    plt.show()
+    plt.show()
+
+
+def read_golist(filepath):
+
+    golist = {}
+    with open(filepath, 'r') as f:
+
+        for line in f:
+            if line.startswith('id:'):
+                GOid = line.lstrip('id:').strip()
+                golist[GOid] = {}
+            if line.startswith('name:'):
+                GOname = line.lstrip('name:').strip()
+                golist[GOid]['description'] = GOname
+            if line.startswith('namespace:'):
+                GOonto = line.lstrip('namespace:').strip()
+                golist[GOid]['ontology'] = GOonto
+
+    return golist
+
+
+def assign_go(file_path, golist):
+
+    data = pd.read_csv(file_path)
+    data['Description'] = data['ID'].apply(lambda x: golist.get(x, {}).get('description', 'NA(Obsolete)'))
+    data['Ontology'] = data['ID'].apply(lambda x: golist.get(x, {}).get('ontology', 'NA(Obsolete)'))
+    dirpath = os.path.dirname(file_path)
+    data.to_csv(f'{dirpath}/GOanno.csv', index=False)

ui_1.py

@@ -171,7 +171,7 @@ def eventFilter(self, source, event):
                 source.setText(file_path)
                 return True
 
-        if source in [self.cds_input_text, self.vol_file, self.enrich_file]:
+        elif source in [self.cds_input_text, self.vol_file, self.enrich_file]:
             if event.type() == QEvent.DragEnter:
                 if event.mimeData().hasUrls():
                     event.acceptProposedAction()

ui_2.py

@@ -9,7 +9,7 @@
 from PyQt5.QtCore import QUrl, QEvent, QTimer
 from PyQt5.QtGui import QDesktopServices, QIcon
 from ui_1 import MyApp
-from plot import plot_GOem, plot_GOem_classify
+from plot import plot_GOem, plot_GOem_classify, read_golist, assign_go
 import sys
 import os
 import urllib.request
@@ -22,12 +22,14 @@ def __init__(self):
         self.enrich_file.installEventFilter(self)
         self.gtl_dl.clicked.connect(self.gtlfile_dl)
         self.plot_bu_2.clicked.connect(self.GOem_plot)
+        self.gef_c.clicked.connect(self.GOem_anno)
+
 
     def gtlfile_dl(self):
 
         gtl_url = 'http://current.geneontology.org/ontology/go-basic.obo'
         gtl_fn = 'go_term.list'
-        if os.path.exists(gtl_fn):
+        if os.path.exists(gtl_fn) or os.path.exists('go-basic.obo'):
             self.show_message_dialog("文件已存在，下载终止")
             return
         try:
@@ -58,6 +60,28 @@ def GOem_plot(self):
             except Exception as e:
                 self.show_message_dialog(f"发生错误：{e}\n请检查文件")
 
+    def GOem_anno(self):
+
+        fp = self.enrich_file.toPlainText()
+        if fp == '':
+            self.show_message_dialog("请输入文件路径。")
+            return
+
+        if os.path.exists('go_term.list'):
+            glfile = 'go_term.list'
+        elif os.path.exists('go-basic.obo'):
+            glfile = 'go-basic.obo'
+        else:
+            self.show_message_dialog("不存在 GO list 文件，请下载。")
+            return
+
+        golist = read_golist(glfile)
+        if golist == {}:
+            self.show_message_dialog("GO list 文件存在错误，请检查。")
+            return
+
+        assign_go(fp, golist)
+        self.show_message_dialog("请完成注释，请见 GOanno.csv。")
 
 if __name__ == "__main__":
     app = QApplication(sys.argv)