Add contaminants.fasta

fcyu · Jan 29, 2018 · 9b48fcb · 9b48fcb
1 parent 8398aa4
commit 9b48fcb
Show file tree

Hide file tree

Showing 3 changed files with 2,771 additions and 40 deletions.
diff --git a/src/main/java/proteomics/Index/BuildIndex.java b/src/main/java/proteomics/Index/BuildIndex.java
@@ -7,6 +7,7 @@
 import proteomics.TheoSeq.MassTool;
 import proteomics.Types.*;
 
+import java.io.IOException;
 import java.util.*;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
@@ -30,7 +31,7 @@ public class BuildIndex {
     private Map<String, Set<String>> seqProMap;
     private final float ms1_bin_size;
 
-    public BuildIndex(Map<String, String> parameter_map) {
+    public BuildIndex(Map<String, String> parameter_map) throws IOException {
         // initialize parameters
         int min_chain_length = Integer.valueOf(parameter_map.get("min_chain_length")) + 2; // n and c are counted in the sequence
         int max_chain_length = Integer.valueOf(parameter_map.get("max_chain_length")) + 2; // n and c are counted in the sequence
@@ -93,8 +94,11 @@ public BuildIndex(Map<String, String> parameter_map) {
 
         // read protein database
         DbTool db_tool_obj = new DbTool(db_path, parameter_map.get("database_type"));
-        Map<String, String> pro_seq_map = db_tool_obj.getProSeqMap();
-        pro_annotate_map = db_tool_obj.getProAnnotateMap();
+        DbTool contaminantsDb = new DbTool(null, "contaminants");
+        Map<String, String> pro_seq_map = contaminantsDb.getProSeqMap();
+        pro_seq_map.putAll(db_tool_obj.getProSeqMap()); // using the target sequence to replace contaminant sequence if there is conflict.
+        pro_annotate_map = contaminantsDb.getProAnnotateMap();
+        pro_annotate_map.putAll(db_tool_obj.getProAnnotateMap()); // using the target sequence to replace contaminant sequence if there is conflict.
 
         // define a new MassTool object
         mass_tool_obj = new MassTool(missed_cleavage, fix_mod_map, "KR", "P", mz_bin_size, one_minus_bin_offset);

diff --git a/src/main/java/proteomics/TheoSeq/DbTool.java b/src/main/java/proteomics/TheoSeq/DbTool.java
@@ -14,7 +14,7 @@ public class DbTool {
     private Map<String, String> pro_seq_map = new HashMap<>();
     private Map<String, String> pro_annotate_map = new HashMap<>();
 
-    public DbTool(String db_name, String databaseType) {
+    public DbTool(String db_name, String databaseType) throws IOException {
         String id = "";
         String annotate;
         StringBuilder seq = new StringBuilder(99999);
@@ -26,52 +26,55 @@ public DbTool(String db_name, String databaseType) {
             header_pattern = Pattern.compile("^>([^\\s]+)[\\s|]+(.+)$");
         } else if (databaseType.contentEquals("UniProt") || databaseType.contentEquals("SwissProt")) {
             header_pattern = Pattern.compile("^>[^|]+\\|(.+)\\|(.+)$");
+        } else if (databaseType.contentEquals("contaminants")) {
+            header_pattern = Pattern.compile("^>([^ ]+) (.+)$");
         } else if (databaseType.contentEquals("Others")) {
             header_pattern = Pattern.compile("^>(.+)$");
-        }
-        else {
+        } else {
             header_pattern = null;
             logger.error("Incorrect database type ({}) in the parameter file.", databaseType);
             System.exit(1);
         }
 
-        try (BufferedReader db_reader = new BufferedReader(new FileReader(db_name))) {
-            String line;
-            while ((line = db_reader.readLine()) != null) {
-                line = line.trim();
-                Matcher head_matcher = header_pattern.matcher(line);
-                if (head_matcher.matches()) {
-                    // This line is a header
-                    if (!new_pro) {
-                        // This isn't the first protein
-                        pro_seq_map.put(id, seq.toString());
-                    }
-                    id = head_matcher.group(1).trim();
-                    if (databaseType.contentEquals("Others")) {
-                        annotate = id;
-                    } else {
-                        annotate = head_matcher.group(2).trim();
-                    }
-                    pro_annotate_map.put(id, annotate);
-                    new_pro = true;
-                } else if (!line.isEmpty()) {
-                    // This line is a body
-                    if (new_pro) {
-                        seq = new StringBuilder(99999);
-                        seq.append(line);
-                        new_pro = false;
-                    } else {
-                        seq.append(line);
-                    }
+        BufferedReader db_reader;
+        if (databaseType.contentEquals("contaminants")) {
+            InputStream inputStream = getClass().getClassLoader().getResourceAsStream("contaminants.fasta");
+            db_reader = new BufferedReader(new InputStreamReader(inputStream));
+        } else {
+            db_reader = new BufferedReader(new FileReader(db_name));
+        }
+        String line;
+        while ((line = db_reader.readLine()) != null) {
+            line = line.trim();
+            Matcher head_matcher = header_pattern.matcher(line);
+            if (head_matcher.matches()) {
+                // This line is a header
+                if (!new_pro) {
+                    // This isn't the first protein
+                    pro_seq_map.put(id, seq.toString());
+                }
+                id = head_matcher.group(1).trim();
+                if (databaseType.contentEquals("Others")) {
+                    annotate = id;
+                } else {
+                    annotate = head_matcher.group(2).trim();
+                }
+                pro_annotate_map.put(id, annotate);
+                new_pro = true;
+            } else if (!line.isEmpty()) {
+                // This line is a body
+                if (new_pro) {
+                    seq = new StringBuilder(99999);
+                    seq.append(line);
+                    new_pro = false;
+                } else {
+                    seq.append(line);
                 }
             }
-            // Last protein
-            pro_seq_map.put(id, seq.toString());
-        } catch (IOException | PatternSyntaxException ex) {
-            logger.error(ex.toString());
-            ex.printStackTrace();
-            System.exit(1);
         }
+        db_reader.close();
+        // Last protein
+        pro_seq_map.put(id, seq.toString());
     }
 
     public Map<String, String> getProSeqMap() {