Squashed initial commit

2024-09-10 13:47:29 -04:00
commit 8ebb6ad265
6221 changed files with 2512206 additions and 0 deletions
--- a/qhtcp-workflow/apps/java/weka-clustering/src/SGD2AttrTable.java
+++ b/qhtcp-workflow/apps/java/weka-clustering/src/SGD2AttrTable.java
@@ -0,0 +1,155 @@
+
+import java.io.BufferedReader;
+import java.io.File;
+import java.io.FileOutputStream;
+import java.io.FileReader;
+import java.io.IOException;
+import java.io.OutputStreamWriter;
+import java.util.HashMap;
+import java.util.Iterator;
+
+
+/*
+ * This program starts by creating an intermediate table and then will load the function from Dr. Brett McKinney to create the attribute table.
+ */
+
+/**
+ *
+ * @author DTian
+ */
+public class SGD2AttrTable {
+
+    public void createIntermediateTable(String inputFile, String outputFile) {
+        HashMap geneToGODict  = new HashMap();
+         try {
+
+            FileReader fr = new FileReader(inputFile);
+            BufferedReader br = new BufferedReader(fr);
+
+            // strRow is used to read line from file(skip first row)
+            String strRow = br.readLine();
+
+            // The while loop read the data from data file to vvf
+            while ((strRow = br.readLine()) != null) {
+
+                // Check: skip the line if it is a comment line
+                if (strRow.trim().charAt(0) != 'S' ) {
+                    continue;
+                }
+                String [] strArray = strRow.trim().split("\\t");
+                String key = toKey(strArray[10].toUpperCase());
+                if (key.compareToIgnoreCase("") == 0) {
+                    continue;
+                }
+                String value = toValue(strArray[4]);
+                if (geneToGODict.containsKey(key)) {
+                    geneToGODict.put(key, geneToGODict.get(key)+ "\t" + value);
+                } else {
+                    geneToGODict.put(key, value);
+                }
+            }
+
+            br.close();
+            fr.close();
+
+            // Write to output file
+            FileOutputStream stream; // provides file access
+            OutputStreamWriter writer; // writes to the file
+            stream = new FileOutputStream(new File(outputFile), true);
+            writer = new OutputStreamWriter(stream);
+            Iterator it = geneToGODict.keySet().iterator();
+            while(it.hasNext()){
+                String key = it.next().toString();
+                String value = geneToGODict.get(key).toString();
+                writer.write(key + "\t" + value + "\n");
+            }
+            writer.flush();
+            writer.close();
+            stream.close();
+        } catch (IOException e) {
+            // Catch possible io errors from readLine()
+            System.out.println("IOException error in  'class SGD2AttrTable, method createIntermediateTable'");
+        }
+    }
+
+    public void createAttrTable(String intermediaFile, String outputFile){
+        HashMap geneToGODict  = new HashMap();
+         try {
+
+            FileReader fr = new FileReader(intermediaFile);
+            BufferedReader br = new BufferedReader(fr);
+
+            // strRow is used to read line from file(skip first row)
+            String strRow = br.readLine();
+
+            // The while loop read the data from data file to vvf
+            while ((strRow = br.readLine()) != null) {
+
+                //check: skip the line if it is a comment line
+                if (strRow.trim().charAt(0) != 'S' ) {
+                    continue;
+                }
+                String [] strArray = strRow.trim().split("\\t");
+                String key = toKey(strArray[10].toUpperCase());
+                if (key.compareToIgnoreCase("") == 0) {
+                    continue;
+                }
+                String value = toValue(strArray[4]);
+                if (geneToGODict.containsKey(key)) {
+                    geneToGODict.put(key, geneToGODict.get(key)+ "\t" + value);
+                } else {
+                    geneToGODict.put(key, value);
+                }
+            }
+
+            br.close();
+            fr.close();
+
+            // Write to output file
+            FileOutputStream stream; // provides file access
+            OutputStreamWriter writer; // writes to the file
+            stream = new FileOutputStream(new File(outputFile), true);
+            writer = new OutputStreamWriter(stream);
+            Iterator it = geneToGODict.keySet().iterator();
+            while (it.hasNext()) {
+                String key = it.next().toString();
+                String value = geneToGODict.get(key).toString();
+                writer.write(key + "\t" + value + "\n");
+            }
+            writer.flush();
+            writer.close();
+            stream.close();
+        } catch (IOException e) {
+            // Catch possible io errors from readLine()
+            System.out.println("IOException error in  'class SGD2AttrTable, method createIntermediateTable'");
+        }
+    }
+
+    /**
+     * 
+     * @param raw  the string need to be get rid of the "GO:0s"
+     * @return the string without "GO:00"
+     */
+    private String toValue(String raw) {
+        String result = raw.toUpperCase(); //raw should be like: "GO:0005739"
+        // Delete "GO:"
+        result = result.substring(3);
+        // Delete "lead zeros"
+        while (result.charAt(0) == '0') {
+            result =result.substring(1);
+        }
+        return result;
+    }
+
+    private String toKey(String raw) {
+        String result = raw.toUpperCase(); // raw should be like: "GO:0005739"
+        // Find the '|'
+        int end = result.indexOf('|');
+        // Get the sub string
+        if (end < 0) {
+            return result;
+        } else {
+            return result.substring(0, end);
+        }
+    }
+}