diff --git a/pom.xml b/pom.xml
index f1b69e7..bb85f78 100644
--- a/pom.xml
+++ b/pom.xml
@@ -4,7 +4,7 @@
hk.ust.bioinformatics
ECL2
- 2.1.4-dev-1c9b46a
+ 2.1.4-dev-201705262001
jar
ECL2
diff --git a/src/main/java/proteomics/ECL2.java b/src/main/java/proteomics/ECL2.java
index 1618075..489eb1a 100644
--- a/src/main/java/proteomics/ECL2.java
+++ b/src/main/java/proteomics/ECL2.java
@@ -25,7 +25,7 @@ public class ECL2 {
public static final boolean flankingPeaks = true;
private static final Logger logger = LoggerFactory.getLogger(ECL2.class);
- public static final String version = "2.1.4-dev-1c9b46a";
+ public static final String version = "2.1.4-dev-201705262001";
public static boolean debug;
public static boolean dev;
diff --git a/src/main/java/proteomics/Index/BuildIndex.java b/src/main/java/proteomics/Index/BuildIndex.java
index 8407448..ce5473e 100644
--- a/src/main/java/proteomics/Index/BuildIndex.java
+++ b/src/main/java/proteomics/Index/BuildIndex.java
@@ -6,6 +6,7 @@
import proteomics.TheoSeq.DbTool;
import proteomics.TheoSeq.MassTool;
import proteomics.Types.AA;
+import proteomics.Types.BinaryModParam;
import proteomics.Types.ChainEntry;
import proteomics.Types.VarModParam;
@@ -16,7 +17,7 @@
public class BuildIndex {
private static final Logger logger = LoggerFactory.getLogger(BuildIndex.class);
- private static final Pattern varModParamPattern = Pattern.compile("([0-9.-]+)\\s+([A-Znc]+)");
+ private static final Pattern varModParamPattern = Pattern.compile("([0-9.-]+)\\s+([A-Znc]+)\\s+([01])");
private static final int globalVarModMaxNum = 5; // Do not change this value. Otherwise, change generateLocalIdxModMassMap accordingly.
private static final float varModMassResolution = 0.01f;
@@ -94,26 +95,27 @@ public BuildIndex(Map parameter_map) {
seqProMap = buildSeqProMap(pro_seq_map, min_chain_length, max_chain_length);
// read var mods
- Set varModParams = new HashSet<>();
+ Set varModParamSet = new HashSet<>();
+ Set binaryModParamSet = new HashSet<>();
for (String k : parameter_map.keySet()) {
if (k.contentEquals("var_mod1")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod2")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod3")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod4")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod5")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod6")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod7")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod8")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
} else if (k.contentEquals("var_mod9")) {
- varModParams.addAll(getVarModParams(parameter_map.get(k)));
+ getVarModParams(parameter_map.get(k), varModParamSet, binaryModParamSet);
}
}
@@ -144,7 +146,7 @@ public BuildIndex(Map parameter_map) {
}
// mod containing
- Set varSeqSet = generateModSeq(seq, linkSiteSet, varModParams, varModMaxNum);
+ Set varSeqSet = generateModSeq(seq, linkSiteSet, varModParamSet, binaryModParamSet, varModMaxNum);
for (String varSeq : varSeqSet) {
linkSiteSet = getLinkSiteSet(varSeq, proteinNTerm, proteinCTerm);
if (!linkSiteSet.isEmpty()) {
@@ -290,38 +292,121 @@ private Map> buildSeqProMap(Map pro_seq_map,
return seq_pro_map;
}
- private Set generateModSeq(String seq, Set modFreeListSites, Set varModParams, int varModMaxNum) { // todo: check
- // get all locations' var lists
- Map> idxModMassMap = new HashMap<>();
- for (int i = 0; i < seq.length(); ++i) {
- char aa = seq.charAt(i);
- for (VarModParam varModParam : varModParams) {
- if (varModParam.aa == aa) {
- if (idxModMassMap.containsKey(i)) {
- idxModMassMap.get(i).add(varModParam.modMass);
- } else {
- List temp = new LinkedList<>();
- temp.add(varModParam.modMass);
- idxModMassMap.put(i, temp);
+ private Set generateModSeq(String seq, Set modFreeListSites, Set varModParamSet, Set binaryModParamSet, int varModMaxNum) { // todo: check
+ Set varSeqSet = new HashSet<>();
+ for (short linkSite : modFreeListSites) {
+ // has binary mod
+ for (BinaryModParam binaryModParam : binaryModParamSet) {
+ // get all locations having binary mod
+ Map> idxBinaryModMassMap = new HashMap<>();
+ for (int i = 0; i < seq.length(); ++i) {
+ if (i != linkSite) {
+ String aa = seq.substring(i, i + 1);
+ if (binaryModParam.aas.contains(aa)) {
+ List tempList = new LinkedList<>();
+ tempList.add(binaryModParam.modMass);
+ idxBinaryModMassMap.put(i, tempList);
+ }
+ }
+ }
+ if (!idxBinaryModMassMap.isEmpty()) {
+ // generate a sequence only containing the binary mod
+ StringBuilder sb = new StringBuilder(seq.length() * 10);
+ for (int i = 0; i < seq.length(); ++i) {
+ sb.append(seq.substring(i, i + 1));
+ if (idxBinaryModMassMap.containsKey(i)) {
+ sb.append(String.format("[%.2f]", idxBinaryModMassMap.get(i).get(0)));
+ }
+ }
+ varSeqSet.add(sb.toString());
+
+ if (idxBinaryModMassMap.size() < varModMaxNum) {
+ // generate sequences containing the binary mod and additional var mod
+ // get all locations having var mods
+ Map> idxVarModMassMap = new HashMap<>();
+ for (int i = 0; i < seq.length(); ++i) {
+ if (i != linkSite) {
+ if (!idxBinaryModMassMap.containsKey(i)) {
+ char aa = seq.charAt(i);
+ for (VarModParam varModParam : varModParamSet) {
+ if (varModParam.aa == aa) {
+ if (idxVarModMassMap.containsKey(i)) {
+ idxVarModMassMap.get(i).add(varModParam.modMass);
+ } else {
+ List temp = new LinkedList<>();
+ temp.add(varModParam.modMass);
+ idxVarModMassMap.put(i, temp);
+ }
+ }
+ }
+ }
+ }
+ }
+ if (!idxVarModMassMap.isEmpty()) {
+ // generate var containing sequences
+ Map> idxBinaryVarModMassMap = new HashMap<>();
+ idxBinaryVarModMassMap.putAll(idxBinaryModMassMap);
+ idxBinaryVarModMassMap.putAll(idxVarModMassMap);
+ Integer[] allIdxArray = idxVarModMassMap.keySet().toArray(new Integer[idxVarModMassMap.size()]);
+ Arrays.sort(allIdxArray);
+ for (int i = 1; i <= Math.min(varModMaxNum - idxBinaryModMassMap.size(), idxVarModMassMap.size()); ++i) {
+ List idxCombinationList = generateIdxCombinations(allIdxArray, i);
+ Set varSetSubSet = new HashSet<>();
+ for (int[] idxCombination : idxCombinationList) {
+ int[] allIdxCombination = new int[idxCombination.length + idxBinaryModMassMap.size()];
+ int j = 0;
+ for (int idx : idxBinaryModMassMap.keySet()) {
+ allIdxCombination[j] = idx;
+ ++j;
+ }
+ for (int k = 0; k < idxCombination.length; ++k) {
+ allIdxCombination[j + k] = idxCombination[k];
+ }
+ Arrays.sort(allIdxCombination);
+ varSetSubSet.addAll(generateModSeqSub(seq, allIdxCombination, idxBinaryVarModMassMap));
+ }
+ if (!varSetSubSet.isEmpty()) {
+ varSeqSet.addAll(checkKCTermMod(varSetSubSet)); // eliminate those sequence that the middle amino acids having the same mod mass and the n-term and the first amino acid or the c-term and the last amino acid have the same mod mass.
+ }
+ }
+ }
}
}
}
- }
- // generate var containing sequences
- Set varSeqSet = new HashSet<>();
- Integer[] allIdxArray = idxModMassMap.keySet().toArray(new Integer[idxModMassMap.size()]);
- Arrays.sort(allIdxArray);
- for (int i = 1; i <= Math.min(varModMaxNum, idxModMassMap.size()); ++i) {
- List idxCombinationList = generateIdxCombinations(allIdxArray, i);
- Set varSetSubSet = new HashSet<>();
- for (int[] idxCombination : idxCombinationList) {
- if (stillHasLinkSite(idxCombination, modFreeListSites)) {
- varSetSubSet.addAll(generateModSeqSub(seq, idxCombination, idxModMassMap));
+ // does not have binary mod
+ // get all locations' var lists
+ Map> idxVarModMassMap = new HashMap<>();
+ for (int i = 0; i < seq.length(); ++i) {
+ if (i != linkSite) {
+ char aa = seq.charAt(i);
+ for (VarModParam varModParam : varModParamSet) {
+ if (varModParam.aa == aa) {
+ if (idxVarModMassMap.containsKey(i)) {
+ idxVarModMassMap.get(i).add(varModParam.modMass);
+ } else {
+ List temp = new LinkedList<>();
+ temp.add(varModParam.modMass);
+ idxVarModMassMap.put(i, temp);
+ }
+ }
+ }
}
}
- if (!varSetSubSet.isEmpty()) {
- varSeqSet.addAll(checkKCTermMod(varSetSubSet)); // eliminate those sequence that the middle amino acids having the same mod mass and the n-term and the first amino acid or the c-term and the last amino acid have the same mod mass.
+ if (!idxVarModMassMap.isEmpty()) {
+ // generate var containing sequences
+ Integer[] allIdxArray = idxVarModMassMap.keySet().toArray(new Integer[idxVarModMassMap.size()]);
+ Arrays.sort(allIdxArray);
+ for (int i = 1; i <= Math.min(varModMaxNum, idxVarModMassMap.size()); ++i) {
+ List idxCombinationList = generateIdxCombinations(allIdxArray, i);
+ Set varSetSubSet = new HashSet<>();
+ for (int[] idxCombination : idxCombinationList) {
+ varSetSubSet.addAll(generateModSeqSub(seq, idxCombination, idxVarModMassMap));
+ }
+ if (!varSetSubSet.isEmpty()) {
+ varSeqSet.addAll(checkKCTermMod(varSetSubSet)); // eliminate those sequence that the middle amino acids having the same mod mass and the n-term and the first amino acid or the c-term and the last amino acid have the same mod mass.
+ }
+ }
}
}
@@ -344,15 +429,6 @@ private List generateIdxCombinations(Integer[] allIdxArray, int num) {
return outputList;
}
- private boolean stillHasLinkSite(int[] idxCombination, Set modFreeLinkSites) {
- for (int modFreeLinkSite : modFreeLinkSites) {
- if (Arrays.binarySearch(idxCombination, modFreeLinkSite) < 0) {
- return true;
- }
- }
- return false;
- }
-
private Set generateModSeqSub(String seq, int[] idxCombination, Map> idxModMassMap) {
List> localIdxModMassMaps = generateLocalIdxModMassMap(idxCombination, idxModMassMap);
@@ -493,24 +569,24 @@ private Set getLinkSiteSet(String seq, boolean n_term, boolean c_term) {
return output;
}
- private Set getVarModParams(String v) {
- Set varModParams = new HashSet<>();
-
+ private void getVarModParams(String v, Set varModParamSet, Set binaryModParamSet) {
Matcher varModMatcher = varModParamPattern.matcher(v);
if (varModMatcher.matches()) {
float modMass = Float.valueOf(varModMatcher.group(1));
String aas = varModMatcher.group(2);
- if (Math.abs(modMass) < varModMassResolution) {
- return varModParams;
- }
- for (int i = 0; i < aas.length(); ++i) {
- varModParams.add(new VarModParam(modMass, aas.charAt(i)));
+ boolean isBinary = varModMatcher.group(3).contentEquals("1");
+ if (Math.abs(modMass) > varModMassResolution) {
+ if (isBinary) {
+ binaryModParamSet.add(new BinaryModParam(modMass, aas));
+ } else {
+ for (int i = 0; i < aas.length(); ++i) {
+ varModParamSet.add(new VarModParam(modMass, aas.charAt(i)));
+ }
+ }
}
} else {
logger.error("Cannot parse variable modification parameter from {}.", v);
System.exit(1);
}
-
- return varModParams;
}
}
diff --git a/src/main/java/proteomics/Types/BinaryModParam.java b/src/main/java/proteomics/Types/BinaryModParam.java
new file mode 100644
index 0000000..adebfd5
--- /dev/null
+++ b/src/main/java/proteomics/Types/BinaryModParam.java
@@ -0,0 +1,31 @@
+package proteomics.Types;
+
+
+public class BinaryModParam {
+ public final float modMass;
+ public final String aas;
+ public final String toString;
+
+ public BinaryModParam(float modMass, String aas) {
+ this.modMass = modMass;
+ this.aas = aas;
+ toString = modMass + "@" + aas + "(binary)";
+ }
+
+ public String toString() {
+ return toString;
+ }
+
+ public int hashCode() {
+ return toString.hashCode();
+ }
+
+ public boolean equals(Object other) {
+ if (other instanceof BinaryModParam) {
+ BinaryModParam temp = (BinaryModParam) other;
+ return (Math.abs(temp.modMass - modMass) <= 0.01) && (temp.aas.contentEquals(aas));
+ } else {
+ return false;
+ }
+ }
+}
diff --git a/src/main/java/proteomics/Types/BinaryVarMod.java b/src/main/java/proteomics/Types/BinaryVarMod.java
deleted file mode 100644
index c4fee5d..0000000
--- a/src/main/java/proteomics/Types/BinaryVarMod.java
+++ /dev/null
@@ -1,31 +0,0 @@
-package proteomics.Types;
-
-public class BinaryVarMod {
-
- public final String site;
- public final float mod_mass;
- private final String to_string;
-
- public BinaryVarMod(String site, float mod_mass) {
- this.site = site;
- this.mod_mass = mod_mass;
- to_string = String.format("%.2f@%s", mod_mass, site);
- }
-
- public boolean equals(Object other) {
- if (other instanceof BinaryVarMod) {
- BinaryVarMod temp = (BinaryVarMod) other;
- return ((temp.site.contentEquals(site)) && (temp.mod_mass == mod_mass));
- } else {
- return false;
- }
- }
-
- public String toString() {
- return to_string;
- }
-
- public int hashCode() {
- return to_string.hashCode();
- }
-}
diff --git a/src/main/resources/parameter.def b/src/main/resources/parameter.def
index a7bd11a..5b87040 100644
--- a/src/main/resources/parameter.def
+++ b/src/main/resources/parameter.def
@@ -1,5 +1,5 @@
-# 2.1.4-dev
-# The first line is the parameter file version. Don't change it.
+# 2.1.4-dev-201705262001
+# The first line is the parameter file version. Do not change it.
thread_num = 0
debug = 0
dev = 0
@@ -29,16 +29,18 @@ mz_bin_offset = 0
cl_mass = 138.0680796
# Var modification
-# format:
-var_mod1 = 0.0 X
-var_mod2 = 0.0 X
-var_mod3 = 0.0 X
-var_mod4 = 0.0 X
-var_mod5 = 0.0 X
-var_mod6 = 0.0 X
-var_mod7 = 0.0 X
-var_mod8 = 0.0 X
-var_mod9 = 0.0 X
+# format:
+# == 0/1
+# binary modification is mutual exclusion with each other
+var_mod1 = 15.99 M 0
+var_mod2 = 0.0 X 0
+var_mod3 = 0.0 X 0
+var_mod4 = 0.0 X 0
+var_mod5 = 0.0 X 0
+var_mod6 = 0.0 X 0
+var_mod7 = 0.0 X 0
+var_mod8 = 0.0 X 0
+var_mod9 = 0.0 X 0
var_mod_max_num = 5 # max number of modified amino acids in a peptide. The max value is 5
# Fix modification