Merge remote-tracking branch 'origin/master' into faster-lookup

ebi-pf-team · Dec 2, 2024 · b103587 · b103587
2 parents ce8deaf + 1f48e66
commit b103587
Show file tree

Hide file tree

Showing 410 changed files with 72,698 additions and 111,636 deletions.
diff --git a/.travis.yml b/.travis.yml
diff --git a/Dockerfile b/Dockerfile
@@ -2,13 +2,13 @@ FROM ubuntu:20.04
 
 LABEL authors="Laise Florentino ([email protected]), Matthias Blum ([email protected])"
 
-ARG VERSION=5.69-101.0
+ARG VERSION
 ENV TZ=Europe/London
 
 RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && \
     echo $TZ > /etc/timezone && \
     apt-get update -y && \
-    apt-get install -y wget python3.8 python2.7 openjdk-11-jre-headless libpcre2-dev libgomp1 perl-doc && \
+    apt-get install -y wget python3.8 openjdk-11-jre-headless libpcre2-dev libgomp1 perl-doc libc6-i386 && \
     ln -s /usr/bin/python3.8 /usr/bin/python3 && ln -s /usr/bin/python3.8 /usr/bin/python
 
 WORKDIR /opt
@@ -17,8 +17,15 @@ RUN wget ftp://ftp.ebi.ac.uk/pub/software/unix/iprscan/5/${VERSION}/alt/interpro
     wget ftp://ftp.ebi.ac.uk/pub/software/unix/iprscan/5/${VERSION}/alt/interproscan-core-${VERSION}.tar.gz.md5 && \
     md5sum -c interproscan-core-${VERSION}.tar.gz.md5 && \
     tar -zxf interproscan-core-${VERSION}.tar.gz && \
-    rm interproscan-core-${VERSION}.tar.gz && \
+    rm interproscan-core-${VERSION}.tar.gz* && \
     mv /opt/interproscan-${VERSION} /opt/interproscan
 
 WORKDIR /opt/interproscan
+
+RUN echo "binary.phobius.pl.path=/opt/interproscan/licensed/phobius/phobius.pl" >> interproscan.properties && \
+    echo "binary.signalp.path=/opt/interproscan/licensed/signalp/signalp" >> interproscan.properties && \
+    echo "signalp.perl.library.dir=/opt/interproscan/licensed/signalp/lib" >> interproscan.properties && \
+    echo "binary.tmhmm.path=/opt/interproscan/licensed/tmhmm/decodeanhmm.Linux_x86_64" >> interproscan.properties && \
+    echo "tmhmm.model.path=/opt/interproscan/licensed/tmhmm/TMHMM2.0.model" >> interproscan.properties
+
 ENTRYPOINT ["bash", "./interproscan.sh"]
diff --git a/README.md b/README.md
@@ -1,7 +1,7 @@
 [![Build](https://github.com/ebi-pf-team/interproscan/actions/workflows/build-test.yml/badge.svg)](https://github.com/ebi-pf-team/interproscan/actions/workflows/build-test.yml)
-[![Docker Image Version (latest semver)](https://img.shields.io/docker/v/interpro/interproscan?label=Docker)](https://hub.docker.com/r/interpro/interproscan)
+[![Docker Image Version (tag)](https://img.shields.io/docker/v/interpro/interproscan/5.72-103.0)](https://hub.docker.com/r/interpro/interproscan)
 
-# interproscan
+# InterProScan
 
 ## What is InterProScan?
 [InterPro](http://www.ebi.ac.uk/interpro/) is a database which integrates together predictive information about proteins' function from a number of partner resources, giving an overview of the families that a protein belongs to and the domains and sites it contains.

diff --git a/core/business/pom.xml b/core/business/pom.xml
@@ -5,13 +5,13 @@
     <parent>
         <groupId>uk.ac.ebi.interpro.scan</groupId>
         <artifactId>interproscan</artifactId>
-        <version>5.69-101.0</version>
+        <version>5.72-103.0</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>
 
     <artifactId>interproscan-business</artifactId>
-    <version>5.69-101.0</version>
+    <version>5.72-103.0</version>
 
     <packaging>jar</packaging>
     <name>InterProScan Business Logic</name>
@@ -35,14 +35,14 @@
         <dependency>
             <groupId>uk.ac.ebi.interpro.scan</groupId>
             <artifactId>interproscan-persistence</artifactId>
-            <version>5.69-101.0</version>
+            <version>5.72-103.0</version>
         </dependency>
 
 
         <dependency>
             <groupId>uk.ac.ebi.interpro.scan</groupId>
             <artifactId>precalc-match-client</artifactId>
-            <version>5.69-101.0</version>
+            <version>5.72-103.0</version>
         </dependency>
 
         <dependency>

diff --git a/core/generic-jpa-dao/pom.xml b/core/generic-jpa-dao/pom.xml
@@ -5,13 +5,13 @@
     <parent>
         <groupId>uk.ac.ebi.interpro.scan</groupId>
         <artifactId>interproscan</artifactId>
-        <version>5.69-101.0</version>
+        <version>5.72-103.0</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>
 
     <artifactId>generic-jpa-dao</artifactId>
-    <version>5.69-101.0</version>
+    <version>5.72-103.0</version>
 
     <packaging>jar</packaging>
 
@@ -37,7 +37,7 @@
         <dependency>
             <groupId>uk.ac.ebi.interpro.scan</groupId>
             <artifactId>interproscan-util</artifactId>
-            <version>5.69-101.0</version>
+            <version>5.72-103.0</version>
         </dependency>
 
         <!-- Hibernate -->

diff --git a/core/io/pom.xml b/core/io/pom.xml
@@ -5,7 +5,7 @@
     <parent>
         <artifactId>interproscan</artifactId>
         <groupId>uk.ac.ebi.interpro.scan</groupId>
-        <version>5.69-101.0</version>
+        <version>5.72-103.0</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>
@@ -14,21 +14,21 @@
     <packaging>jar</packaging>
 
     <name>InterProScan IO</name>
-    <version>5.69-101.0</version>
+    <version>5.72-103.0</version>
     <description>InterProScan IO</description>
     <url>http://www.ebi.ac.uk/interpro/</url>
 
     <dependencies>
         <dependency>
             <groupId>uk.ac.ebi.interpro.scan</groupId>
             <artifactId>interproscan-util</artifactId>
-            <version>5.69-101.0</version>
+            <version>5.72-103.0</version>
         </dependency>
 
         <dependency>
             <groupId>uk.ac.ebi.interpro.scan</groupId>
             <artifactId>interproscan-model</artifactId>
-            <version>5.69-101.0</version>
+            <version>5.72-103.0</version>
         </dependency>
 
         <dependency>

diff --git a/core/io/src/main/java/uk/ac/ebi/interpro/scan/io/match/mobidb/MobiDBMatchParser.java b/core/io/src/main/java/uk/ac/ebi/interpro/scan/io/match/mobidb/MobiDBMatchParser.java
@@ -1,68 +1,39 @@
 package uk.ac.ebi.interpro.scan.io.match.mobidb;
 
-
-import org.apache.logging.log4j.Logger;
-import org.apache.logging.log4j.LogManager;
 import org.springframework.beans.factory.annotation.Required;
 import uk.ac.ebi.interpro.scan.io.ParseException;
-
-import uk.ac.ebi.interpro.scan.io.match.AbstractLineMatchParser;
 import uk.ac.ebi.interpro.scan.io.match.MatchParser;
 import uk.ac.ebi.interpro.scan.model.SignatureLibrary;
 import uk.ac.ebi.interpro.scan.model.raw.MobiDBRawMatch;
-
-import uk.ac.ebi.interpro.scan.model.raw.PfScanRawMatch;
 import uk.ac.ebi.interpro.scan.model.raw.RawProtein;
-import uk.ac.ebi.interpro.scan.util.Utilities;
 
 import java.io.*;
 import java.util.*;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
 /**
- * Parser for the MobiDB output format:
- * <p>
- * >UNIPARC:UPI00000000FC status=active
- * 165 186
- * //
- * >UNIPARC:UPI0000000107 status=active
- * //
- * >UNIPARC:UPI0000000108 status=active
- * 73 94
- * 123 158
- * //
- * >UNIPARC:UPI0000000109 status=active
- * //
+ * Parser for the IDRPred output format:
+ * PROTEIN_A	16	47	-
+ * PROTEIN_A	115	142	-
+ * PROTEIN_A	117	130	Polyampholyte
+ * PROTEIN_B	609	705	-
+ * PROTEIN_B	622	632	Proline-rich
+ * PROTEIN_B	655	675	Polar
+ * PROTEIN_B	696	705	Polyampholyte
+ * PROTEIN_B	792	820	-
+ * PROTEIN_B	808	820	Polar
+ * PROTEIN_B	874	893	-
+ * PROTEIN_B	948	1566	-
+ * PROTEIN_B	963	974	Polyampholyte
  *
- * @author Gift Nuka
- * @version $Id: MobiDBMatchParser.java,v 5.20 2016/10/21 14:01:17 nuka Exp $
- * @since 1.0-SNAPSHOT
+ * @author Gift Nuka, Matthias Blum
  */
 public class MobiDBMatchParser implements MatchParser<MobiDBRawMatch> {
-
-    private static final Logger LOGGER = LogManager.getLogger(MobiDBMatchParser.class.getName());
-
-    private static final String END_OF_RECORD_MARKER = "//";
-
-    private static final char PROTEIN_ID_LINE_START = '>';
-
     private  SignatureLibrary signatureLibrary;
     private  String signatureLibraryRelease;
 
-    private static final Pattern QUERY_LINE_PATTERN
-            = Pattern.compile("^QUERY\\s+(\\S+)\\s+(\\S+)\\s+(\\d+)\\s+(.*)$");
-    private static final Pattern DOMAIN_LINE_PATTERN
-            = Pattern.compile("^(\\S+)\\s+(\\S+)\\s+(\\S+)(.*)$");
-            //s+(.*)$");
-
-    /**
-     * Matches the line with the start and stop coordinates of the disordered region.
-     * Group 1: Start
-     * Group 2: Stop.
-     */
-    private static final Pattern START_STOP_PATTERN = Pattern.compile("^(\\d+)\\s+(\\d+).*$");
-
+    private static final Pattern DOMAIN_LINE_PATTERN = Pattern.compile("^(\\S+)\\s+(\\S+)\\s+(\\S+)(.*)$");
 
     public MobiDBMatchParser() {
         super();
@@ -73,7 +44,6 @@ public MobiDBMatchParser(SignatureLibrary signatureLibrary, String signatureLibr
         this.signatureLibraryRelease = signatureLibraryRelease;
     }
 
-
     public SignatureLibrary getSignatureLibrary() {
         return signatureLibrary;
     }
@@ -91,7 +61,6 @@ public void setSignatureLibraryRelease(String signatureLibraryRelease) {
         this.signatureLibraryRelease = signatureLibraryRelease;
     }
 
-
     public Set<RawProtein<MobiDBRawMatch>> parse(InputStream is) throws IOException, ParseException {
 
         Map<String, RawProtein<MobiDBRawMatch>> matchData = new HashMap<>();
@@ -115,54 +84,29 @@ public Set<RawProtein<MobiDBRawMatch>> parse(InputStream is) throws IOException,
 
     public Set<MobiDBRawMatch> parseFileInput(InputStream is) throws IOException, ParseException {
         Set<MobiDBRawMatch> matches = new HashSet<>();
+
         try (BufferedReader reader = new BufferedReader(new InputStreamReader(is))) {
             String line;
-            String proteinIdentifier;
-            int lineNumber = 0;
-            String definitionLine = "";
-            String sequenceIdentifier = "";
+
             while ((line = reader.readLine()) != null) {
-                lineNumber++;
-                LOGGER.debug("Line: " + line);
-                //id 80	99
-//                Utilities.verboseLog(1100, "Domain line: " + line);
                 Matcher matcher = DOMAIN_LINE_PATTERN.matcher(line.trim());
 
                 if (matcher.matches()) {
-//                    Utilities.verboseLog(1100, "number of groups: " + matcher.groupCount());
-                    sequenceIdentifier = matcher.group(1);
+                    String sequenceIdentifier = matcher.group(1);
                     int locationStart = Integer.parseInt(matcher.group(2));
                     int locationEnd = Integer.parseInt(matcher.group(3));
                     String feature = matcher.group(4).trim();
-                    if (! (feature.isEmpty() || feature == null)) {
-                        String feature2 = feature;
-//                        Utilities.verboseLog(1100, "We have a feature  : " + feature2 + " " +
-//                                sequenceIdentifier + " region: " + locationStart + "-" +  locationEnd);
-                    }else{
+                    if (feature.equals("-")) {
                         feature = "";
                     }
 
-                    //TODO hardcoded accession should be removed
                     matches.add(new MobiDBRawMatch(sequenceIdentifier, "mobidb-lite",
                             SignatureLibrary.MOBIDB_LITE, signatureLibraryRelease,
                             locationStart, locationEnd, feature));
-//                    Utilities.verboseLog(110, "Match  : " + getLastElement(matches));
                 }
             }
         }
-        Utilities.verboseLog(1100, "MobiDB matches size : " + matches.size());
-        return matches;
-    }
 
-    //get  the last item in the set
-    public Object getLastElement(final Collection c) {
-        final Iterator itr = c.iterator();
-        Object lastElement = itr.next();
-        while (itr.hasNext()) {
-            lastElement = itr.next();
-        }
-        return lastElement;
+        return matches;
     }
-
-
 }
diff --git a/core/jms-implementation/pom.xml b/core/jms-implementation/pom.xml
@@ -5,12 +5,12 @@
     <parent>
         <groupId>uk.ac.ebi.interpro.scan</groupId>
         <artifactId>interproscan</artifactId>
-        <version>5.69-101.0</version>
+        <version>5.72-103.0</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>
     <artifactId>jms-implementation</artifactId>
-    <version>5.69-101.0</version>
+    <version>5.72-103.0</version>
 
     <packaging>jar</packaging>
 
@@ -53,7 +53,7 @@
         <dependency>
             <groupId>uk.ac.ebi.interpro.scan</groupId>
             <artifactId>interproscan-management</artifactId>
-            <version>5.69-101.0</version>
+            <version>5.72-103.0</version>
         </dependency>
 
         <dependency>