From 854efb8498c4f4c640adeaff4da6854deca0eefc Mon Sep 17 00:00:00 2001
From: valot <valot@b8ef2a07-7df7-436f-90b9-41648038564b>
Date: Wed, 15 Feb 2012 09:34:32 +0000
Subject: [PATCH] Separated contaminants database filter on separated class
git-svn-id: https://subversion.renater.fr/xtandempipeline/trunk@214 b8ef2a07-7df7-436f-90b9-41648038564b
---
.../class_msms/Identification.java | 3 +-
.../FilterGroupFromFastaDatabase.java | 87 +++++++++++++++++++
.../xtandempipeline/grouping/GroupSet.java | 68 +--------------
3 files changed, 91 insertions(+), 67 deletions(-)
create mode 100644 xtandempipeline/src/fr/inra/pappso/xtandempipeline/filter_print/FilterGroupFromFastaDatabase.java
diff --git a/xtandempipeline/src/fr/inra/pappso/xtandempipeline/class_msms/Identification.java b/xtandempipeline/src/fr/inra/pappso/xtandempipeline/class_msms/Identification.java
index c76790a4f..71840ae57 100644
--- a/xtandempipeline/src/fr/inra/pappso/xtandempipeline/class_msms/Identification.java
+++ b/xtandempipeline/src/fr/inra/pappso/xtandempipeline/class_msms/Identification.java
@@ -6,6 +6,7 @@ import java.util.Collections;
import org.apache.log4j.Logger;
import fr.inra.pappso.xtandempipeline.MsException.MSMSException;
+import fr.inra.pappso.xtandempipeline.filter_print.FilterGroupFromFastaDatabase;
import fr.inra.pappso.xtandempipeline.filter_print.filter_base;
import fr.inra.pappso.xtandempipeline.filter_print.print_base;
import fr.inra.pappso.xtandempipeline.filter_print.swt_table_base;
@@ -118,7 +119,7 @@ public class Identification {
public void remove_contaminant_submatch(File database) {
//TODO Remove group containing prot from this database
- grouping.removeGroupFromDatabase(database);
+ new FilterGroupFromFastaDatabase(database).filter(grouping);
}
public void setProteinListValideToGroupeSet() {
diff --git a/xtandempipeline/src/fr/inra/pappso/xtandempipeline/filter_print/FilterGroupFromFastaDatabase.java b/xtandempipeline/src/fr/inra/pappso/xtandempipeline/filter_print/FilterGroupFromFastaDatabase.java
new file mode 100644
index 000000000..67ea419cb
--- /dev/null
+++ b/xtandempipeline/src/fr/inra/pappso/xtandempipeline/filter_print/FilterGroupFromFastaDatabase.java
@@ -0,0 +1,87 @@
+package fr.inra.pappso.xtandempipeline.filter_print;
+
+import java.io.BufferedReader;
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.InputStreamReader;
+import java.util.HashSet;
+
+import org.apache.log4j.Logger;
+
+import fr.inra.pappso.xtandempipeline.grouping.Group;
+import fr.inra.pappso.xtandempipeline.grouping.GroupSet;
+import fr.inra.pappso.xtandempipeline.grouping.HashSampleScanSetProt;
+import fr.inra.pappso.xtandempipeline.grouping.SubGroup;
+
+public class FilterGroupFromFastaDatabase {
+
+ private static final Logger logger = Logger.getLogger(FilterGroupFromFastaDatabase.class);
+
+ private File database;
+
+ public FilterGroupFromFastaDatabase(File database) {
+ logger.info("Remove Group from Fasta Database "+ database.getName());
+ this.database=database;
+ }
+
+ public void filter (GroupSet grouping) {
+ //Read fasta file to get accession list
+ if(database.exists()){
+ logger.info("Read fasta file : "+database.getName());
+ HashSet<String> access = new HashSet<String>();
+ try{
+ FileInputStream is = new FileInputStream(database);
+ InputStreamReader isr = new InputStreamReader(is);
+ BufferedReader br = new BufferedReader(isr);
+
+ String line;
+ while((line= br.readLine())!=null){
+ line.trim();
+ if(line.length()>0){
+ if(line.startsWith(">")){
+ access.add(line.split(" ")[0].replaceFirst(">", ""));
+ }
+ }
+ }
+ }catch (Exception e) {
+ logger.error(e);
+ }
+ //Search contaminants proteins
+ logger.info("Remove Group in new mode");
+ for (Group group : grouping.getGroupList()) {
+ boolean isconta = false;
+ for (SubGroup sg : group.getSubGroupSet()) {
+ for (HashSampleScanSetProt hashProt : sg
+ .getHashSampleScanSetProtSet()) {
+ if (access.contains(hashProt.getMatch().get_protein_match().get_cle_accession())) {
+ isconta = true;
+ }
+ }
+ }
+ if (isconta)
+ grouping.getHashSetOfGroups().remove(group);
+ }
+ }else{
+ logger.error("fasta file doesnt not exist: " + database.getAbsolutePath());
+ logger.info("Remove Group from Database in old mode");
+ for (Group group : grouping.getGroupList()) {
+ boolean isconta = false;
+ for (SubGroup sg : group.getSubGroupSet()) {
+ for (HashSampleScanSetProt hashProt : sg
+ .getHashSampleScanSetProtSet()) {
+ if (hashProt.getMatch().get_protein_match().getDatabase().getDatabasePath()
+ .contains(database.getName())) {
+ isconta = true;
+ }
+ }
+ }
+ if (isconta)
+ grouping.getHashSetOfGroups().remove(group);
+ }
+ }
+
+ //Numbering after contaminant removing
+ grouping.numbering();
+ }
+
+}
diff --git a/xtandempipeline/src/fr/inra/pappso/xtandempipeline/grouping/GroupSet.java b/xtandempipeline/src/fr/inra/pappso/xtandempipeline/grouping/GroupSet.java
index 6e1d7d64a..a0dfa5d5b 100644
--- a/xtandempipeline/src/fr/inra/pappso/xtandempipeline/grouping/GroupSet.java
+++ b/xtandempipeline/src/fr/inra/pappso/xtandempipeline/grouping/GroupSet.java
@@ -1,9 +1,5 @@
package fr.inra.pappso.xtandempipeline.grouping;
-import java.io.BufferedReader;
-import java.io.File;
-import java.io.FileInputStream;
-import java.io.InputStreamReader;
import java.util.ArrayList;
import java.util.Collections;
import java.util.HashSet;
@@ -31,7 +27,7 @@ public class GroupSet {
this.numbering();
}
- private void numbering() {
+ public void numbering() {
// numerotation
ArrayList<Group> orderedGroupList = new ArrayList<Group>(groupList);
@@ -86,67 +82,7 @@ public class GroupSet {
return orderedGroupList;
}
- public void removeGroupFromDatabase(File database) {
- //Read fasta file to get accession list
- if(database.exists()){
- logger.info("Read fasta file : "+database.getName());
- HashSet<String> access = new HashSet<String>();
- try{
- FileInputStream is = new FileInputStream(database);
- InputStreamReader isr = new InputStreamReader(is);
- BufferedReader br = new BufferedReader(isr);
-
- String line;
- while((line= br.readLine())!=null){
- line.trim();
- if(line.length()>0){
- if(line.startsWith(">")){
- access.add(line.split(" ")[0].replaceFirst(">", ""));
- }
- }
- }
- }catch (Exception e) {
- logger.error(e);
- }
- //Search contaminants proteins
- logger.info("Remove Group from Database in new mode: " + database.getName());
- for (Group group : this.getGroupList()) {
- boolean isconta = false;
- for (SubGroup sg : group.getSubGroupSet()) {
- for (HashSampleScanSetProt hashProt : sg
- .getHashSampleScanSetProtSet()) {
- if (access.contains(hashProt.getMatch().get_protein_match().get_cle_accession())) {
- isconta = true;
- }
- }
- }
- if (isconta)
- this.groupList.remove(group);
- }
- }else{
- logger.error("fasta file doesnt not exist: " + database.getAbsolutePath());
- logger.info("Remove Group from Database in old mode: " + database.getName());
- for (Group group : this.getGroupList()) {
- boolean isconta = false;
- for (SubGroup sg : group.getSubGroupSet()) {
- for (HashSampleScanSetProt hashProt : sg
- .getHashSampleScanSetProtSet()) {
- if (hashProt.getMatch().get_protein_match().getDatabase().getDatabasePath()
- .contains(database.getName())) {
- isconta = true;
- }
- }
- }
- if (isconta)
- this.groupList.remove(group);
- }
- }
-
- //Numbering after contaminant removing
- this.numbering();
- }
-
- protected HashSet<Group> getHashSetOfGroups() {
+ public HashSet<Group> getHashSetOfGroups() {
return this.groupList;
}
--
GitLab