From 58b5a7614ce27a57f87da31c1708ef2b5901e0f4 Mon Sep 17 00:00:00 2001
From: Fengchao <fyuab@connect.ust.hk>
Date: Thu, 7 Jun 2018 20:23:26 +0800
Subject: [PATCH] Support more enzymes.

---
 .../java/proteomics/Index/BuildIndex.java     | 28 +++++++++++++++----
 src/main/resources/parameter.def              | 12 ++++++++
 2 files changed, 34 insertions(+), 6 deletions(-)

diff --git a/src/main/java/proteomics/Index/BuildIndex.java b/src/main/java/proteomics/Index/BuildIndex.java
index e3e38f7..d0d8c25 100644
--- a/src/main/java/proteomics/Index/BuildIndex.java
+++ b/src/main/java/proteomics/Index/BuildIndex.java
@@ -99,7 +99,7 @@ public BuildIndex(Map<String, String> parameter_map) throws Exception {
         }
 
         // define a new MassTool object
-        mass_tool_obj = new MassTool(missed_cleavage, fix_mod_map, "KR", "P", true, mz_bin_size * 0.5, one_minus_bin_offset, "N14");
+        mass_tool_obj = new MassTool(missed_cleavage, fix_mod_map, parameter_map.get("cleavage_site").trim(), parameter_map.get("protection_site").trim(), parameter_map.get("is_from_C_term").trim().contentEquals("1"), mz_bin_size * 0.5, one_minus_bin_offset, "N14");
 
         // generate seq_pro_map
         Map<String, boolean[]> seq_term_map = new HashMap<>();
@@ -139,7 +139,7 @@ public BuildIndex(Map<String, String> parameter_map) throws Exception {
             boolean proteinCTerm = seq_term_map.get(seq)[1];
 
             // mod free
-            Set<Short> linkSiteSet = getLinkSiteSet(seq, proteinNTerm, proteinCTerm, linker_type);
+            Set<Short> linkSiteSet = getLinkSiteSet(seq, parameter_map.get("is_from_C_term").trim().contentEquals("1"), parameter_map.get("cleavage_site"), proteinNTerm, proteinCTerm, linker_type);
             if (!linkSiteSet.isEmpty()) {
                 double totalMass = (mass_tool_obj.calResidueMass(seq) + mass_tool_obj.H2O);
                 int bin = massToBin(totalMass);
@@ -548,14 +548,30 @@ private Set<String> checkKCTermMod(Set<String> varSeqSet) { // eliminate those s
         }
     }
 
-    private Set<Short> getLinkSiteSet(String seq, boolean n_term, boolean c_term, short linker_type) {
+    private Set<Short> getLinkSiteSet(String seq, boolean is_from_C_term, String cleavage_site, boolean n_term, boolean c_term, short linker_type) {
         AA[] aa_list = MassTool.seqToAAList(seq);
         Set<Short> output = new HashSet<>(5, 1);
-        for (int i = 1; i < aa_list.length - 2; ++i) {
+        for (int i = 1; i < aa_list.length - 1; ++i) {
             if (linker_type == 1 && aa_list[i].aa == 'K' && (Math.abs(aa_list[i].ptmDeltaMass) < varModMassResolution)) {
-                output.add((short) i);
+                if (is_from_C_term) {
+                    if (i < aa_list.length - 2 || !cleavage_site.contains("K")) {
+                        output.add((short) i);
+                    }
+                } else {
+                    if (i > 1 || !cleavage_site.contains("K")) {
+                        output.add((short) i);
+                    }
+                }
             } else if (linker_type == 2 && aa_list[i].aa == 'C' && (Math.abs(aa_list[i].ptmDeltaMass) < varModMassResolution)) {
-                output.add((short) i);
+                if (is_from_C_term) {
+                    if (i < aa_list.length - 2 || !cleavage_site.contains("C")) {
+                        output.add((short) i);
+                    }
+                } else {
+                    if (i > 1 || !cleavage_site.contains("C")) {
+                        output.add((short) i);
+                    }
+                }
             }
         }
         if (linker_type == 1 && n_term && !output.contains((short) 1) && (Math.abs(aa_list[0].ptmDeltaMass) < varModMassResolution)) {
diff --git a/src/main/resources/parameter.def b/src/main/resources/parameter.def
index c4de686..5051d0d 100644
--- a/src/main/resources/parameter.def
+++ b/src/main/resources/parameter.def
@@ -66,6 +66,18 @@ O = 0
 n = 0
 c = 0
 
+# Enzyme digestion specificities. Only support one enzyme at a time.
+# enzyme name  	 is cut from C-term 	 cleavage site 	 protection site
+Trypsin        	 1                  	 KR            	 P
+# Trypsin/P      1                  	 KR            	 -
+# TrypsinR     	 1                  	 R             	 P
+# LysC         	 1                  	 K             	 P
+# ArgC         	 1                  	 R             	 P
+# Chymotrypsin 	 1                  	 FYW           	 P
+# GluC         	 1                  	 DE            	 P
+# LysN         	 0                  	 K             	 -
+# AspN         	 0                  	 D            	 -
+
 # Advanced parameters.
 # Don't change them unless necessary.
 single_chain_t = 0 # Single chain score threshold.