X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fgui%2Fstructurechooser%2FTDBResultAnalyser.java;h=e817b26bc8d18cfc4bc8934f1b60a66a949b2270;hb=a6f0678764ea06034460236c7a811bbfcad682aa;hp=f73f3978a8fd60dc0bc0670d878e2b571cba1ee8;hpb=e36731274aafe1e930805c19a0f60372c4c6392a;p=jalview.git
diff --git a/src/jalview/gui/structurechooser/TDBResultAnalyser.java b/src/jalview/gui/structurechooser/TDBResultAnalyser.java
index f73f397..e817b26 100644
--- a/src/jalview/gui/structurechooser/TDBResultAnalyser.java
+++ b/src/jalview/gui/structurechooser/TDBResultAnalyser.java
@@ -1,3 +1,23 @@
+/*
+ * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
+ * Copyright (C) $$Year-Rel$$ The Jalview Authors
+ *
+ * This file is part of Jalview.
+ *
+ * Jalview is free software: you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation, either version 3
+ * of the License, or (at your option) any later version.
+ *
+ * Jalview is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
+ * PURPOSE. See the GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with Jalview. If not, see .
+ * The Jalview Authors are detailed in the 'AUTHORS' file.
+ */
package jalview.gui.structurechooser;
import java.util.ArrayList;
@@ -7,6 +27,7 @@ import java.util.Collection;
import java.util.Collections;
import java.util.Comparator;
import java.util.List;
+import java.util.Locale;
import jalview.datamodel.SequenceI;
import jalview.fts.api.FTSData;
@@ -23,8 +44,8 @@ public class TDBResultAnalyser
*/
private static List EXP_CATEGORIES = Arrays
.asList(new String[]
- { "EXPERIMENTALLY DETERMINED", "DEEP LEARNING",
- "TEMPLATE-BASED" });
+ { "EXPERIMENTALLY DETERMINED", "DEEP-LEARNING", "AB-INITIO",
+ "TEMPLATE-BASED", "CONFORMATIONAL ENSEMBLE" });
private SequenceI seq;
@@ -40,28 +61,61 @@ public class TDBResultAnalyser
private int idx_mqual;
+ private int idx_mqualtype;
+
private int idx_resol;
+ /**
+ * selection model
+ */
+ private String filter = null;
+
+ /**
+ * limit to particular source
+ */
+ private String sourceFilter = null;
+
+ private int idx_mprov;
+
public TDBResultAnalyser(SequenceI seq,
Collection collectedResults,
- FTSRestRequest lastTdbRequest)
+ FTSRestRequest lastTdbRequest, String fieldToFilterBy,
+ String string)
{
this.seq = seq;
this.collectedResults = collectedResults;
this.lastTdbRequest = lastTdbRequest;
+ this.filter = fieldToFilterBy;
+ this.sourceFilter = string;
idx_ups = lastTdbRequest.getFieldIndex("Uniprot Start");
idx_upe = lastTdbRequest.getFieldIndex("Uniprot End");
idx_mcat = lastTdbRequest.getFieldIndex("Model Category");
- idx_mqual = lastTdbRequest.getFieldIndex("Qmean");
+ idx_mprov = lastTdbRequest.getFieldIndex("Provider");
+ idx_mqual = lastTdbRequest.getFieldIndex("Confidence");
idx_resol = lastTdbRequest.getFieldIndex("Resolution");
+ idx_mqualtype = lastTdbRequest.getFieldIndex("Confidence Score Type");
}
- private final int scoreCategory(String cat)
+
+ /**
+ * maintain and resolve categories to 'trust order' TODO: change the trust
+ * scheme to something comprehensible.
+ *
+ * @param cat
+ * @return 0 for null cat, less than zero for others
+ */
+ public final int scoreCategory(String cat)
{
- // TODO: make quicker
- int idx = EXP_CATEGORIES.indexOf(cat.toUpperCase());
+ if (cat == null)
+ {
+ return 0;
+ }
+ String upper_cat = cat.toUpperCase(Locale.ROOT);
+ int idx = EXP_CATEGORIES.indexOf(upper_cat);
if (idx == -1)
{
System.out.println("Unknown category: '" + cat + "'");
+ EXP_CATEGORIES.add(upper_cat);
+ idx = EXP_CATEGORIES.size() - 1;
}
return -EXP_CATEGORIES.size() - idx;
}
@@ -79,21 +133,29 @@ public class TDBResultAnalyser
// ignore anything outside the sequence region
for (FTSData row : collectedResults)
{
- int up_s = (Integer) row.getSummaryData()[idx_ups];
- int up_e = (Integer) row.getSummaryData()[idx_upe];
-
- if (seq == row.getSummaryData()[0] && up_e > seq.getStart()
- && up_s < seq.getEnd())
+ if (row.getSummaryData() != null
+ && row.getSummaryData()[idx_ups] != null)
{
- filteredResponse.add(row);
+ int up_s = (Integer) row.getSummaryData()[idx_ups];
+ int up_e = (Integer) row.getSummaryData()[idx_upe];
+ String provider = (String) row.getSummaryData()[idx_mprov];
+ String mcat = (String) row.getSummaryData()[idx_mcat];
+ // this makes sure all new categories are in the score array.
+ int scorecat = scoreCategory(mcat);
+ if (sourceFilter == null || sourceFilter.equals(provider))
+ {
+ if (seq == row.getSummaryData()[0] && up_e > seq.getStart()
+ && up_s < seq.getEnd())
+ {
+ filteredResponse.add(row);
+ }
+ }
}
}
// sort according to decreasing length,
// increasing start
Collections.sort(filteredResponse, new Comparator()
{
-
-
@Override
public int compare(FTSData o1, FTSData o2)
{
@@ -102,9 +164,15 @@ public class TDBResultAnalyser
int o1_s = (Integer) o1data[idx_ups];
int o1_e = (Integer) o1data[idx_upe];
int o1_cat = scoreCategory((String) o1data[idx_mcat]);
+ String o1_prov = ((String) o1data[idx_mprov])
+ .toUpperCase(Locale.ROOT);
int o2_s = (Integer) o2data[idx_ups];
int o2_e = (Integer) o2data[idx_upe];
int o2_cat = scoreCategory((String) o2data[idx_mcat]);
+ String o2_prov = ((String) o2data[idx_mprov])
+ .toUpperCase(Locale.ROOT);
+ String o1_qualtype = (String) o1data[idx_mqualtype],
+ o2_qualtype = (String) o2data[idx_mqualtype];
if (o1_cat == o2_cat)
{
@@ -114,18 +182,58 @@ public class TDBResultAnalyser
int o2_xtent = o2_e - o2_s;
if (o1_xtent == o2_xtent)
{
+ // EXPERIMENTAL DATA ALWAYS TRUMPS MODELS
if (o1_cat == scoreCategory(EXP_CATEGORIES.get(0)))
{
- // experimental structures, so rank on quality
- double o1_res = (Double) o1data[idx_resol];
- double o2_res = (Double) o2data[idx_resol];
- return (o2_res < o1_res) ? 1 : (o2_res == o1_res) ? 0 : -1;
+ if (o1_prov.equals(o2_prov))
+ {
+ if ("PDBE".equals(o1_prov))
+ {
+ if (eitherNull(idx_resol, o1data, o2data))
+ {
+ return nonNullFirst(idx_resol, o1data, o2data);
+ }
+ // experimental structures, so rank on quality
+ double o1_res = (Double) o1data[idx_resol];
+ double o2_res = (Double) o2data[idx_resol];
+ return (o2_res < o1_res) ? 1
+ : (o2_res == o1_res) ? 0 : -1;
+ }
+ else
+ {
+ return 0; // no change in order
+ }
+ }
+ else
+ {
+ // PDBe always ranked above all other experimentally
+ // determined categories
+ return "PDBE".equals(o1_prov) ? -1
+ : "PDBE".equals(o2_prov) ? 1 : 0;
+ }
}
else
{
- // models, so rank on qmean
- float o1_mq = (Float) o1data[idx_mqual];
- float o2_mq = (Float) o2data[idx_mqual];
+ // RANK ON QUALITY - DOWNRANK THOSE WITH NO QUALITY MEASURE
+ if (eitherNull(idx_mqualtype, o1data, o2data))
+ {
+ return nonNullFirst(idx_mqualtype, o1data, o2data);
+ }
+ // ONLY COMPARE LIKE QUALITY SCORES
+ if (!o1_qualtype.equals(o2_qualtype))
+ {
+ // prefer LDDT measure over others
+ return "pLDDT".equals(o1_prov) ? -1
+ : "pLDDT".equals(o2_prov) ? 1 : 0;
+ }
+ // OR NO VALUE FOR THE QUALITY
+ if (eitherNull(idx_mqual, o1data, o2data))
+ {
+ return nonNullFirst(idx_mqual, o1data, o2data);
+ }
+ // models, so rank on qmean - b
+ double o1_mq = (Double) o1data[idx_mqual];
+ double o2_mq = (Double) o2data[idx_mqual];
return (o2_mq < o1_mq) ? 1 : (o2_mq == o1_mq) ? 0 : -1;
}
}
@@ -141,10 +249,29 @@ public class TDBResultAnalyser
}
else
{
+ // if both are not experimental, then favour alphafold
+ if (o2_cat > 0 && o1_cat > 0)
+ {
+ return "ALPHAFOLD DB".equals(o1_prov) ? -1
+ : "ALPHAFOLD DB".equals(o2_prov) ? 1 : 0;
+ }
return o2_cat - o1_cat;
}
}
+ private int nonNullFirst(int idx_resol, Object[] o1data,
+ Object[] o2data)
+ {
+ return o1data[idx_resol] == o2data[idx_resol] ? 0
+ : o1data[idx_resol] != null ? -1 : 1;
+ }
+
+ private boolean eitherNull(int idx_resol, Object[] o1data,
+ Object[] o2data)
+ {
+ return (o1data[idx_resol] == null || o2data[idx_resol] == null);
+ }
+
@Override
public boolean equals(Object obj)
{
@@ -155,48 +282,64 @@ public class TDBResultAnalyser
}
/**
- * return list of structures to be marked as selected for this sequence according to given criteria
- * @param filteredStructures - sorted, filtered structures from getFilteredResponse
+ * return list of structures to be marked as selected for this sequence
+ * according to given criteria
+ *
+ * @param filteredStructures
+ * - sorted, filtered structures from getFilteredResponse
*
*/
public List selectStructures(List filteredStructures)
{
List selected = new ArrayList();
BitSet cover = new BitSet();
- cover.set(seq.getStart(),seq.getEnd());
+ cover.set(seq.getStart(), seq.getEnd());
// walk down the list of structures, selecting some to add to selected
- for (FTSData structure:filteredStructures)
+ // TODO: could do simple DP - double loop to select largest number of
+ // structures covering largest number of sites
+ for (FTSData structure : filteredStructures)
{
- Object[] odata=structure.getSummaryData();
+ Object[] odata = structure.getSummaryData();
int o1_s = (Integer) odata[idx_ups];
int o1_e = (Integer) odata[idx_upe];
int o1_cat = scoreCategory((String) odata[idx_mcat]);
BitSet scover = new BitSet();
// measure intersection
- scover.set(o1_s,o1_e);
+ scover.set(o1_s, o1_e);
scover.and(cover);
- if (scover.cardinality()>4)
+ if (scover.cardinality() > 4)
{
selected.add(structure);
// clear the range covered by this structure
- cover.andNot(scover);
+ cover.andNot(scover);
}
}
- // final step is to sort on length - this might help the superposition process
- Collections.sort(selected,new Comparator()
+ if (selected.size() == 0)
+ {
+ return selected;
+ }
+ // final step is to sort on length - this might help the superposition
+ // process
+ Collections.sort(selected, new Comparator()
{
@Override
public int compare(FTSData o1, FTSData o2)
{
Object[] o1data = o1.getSummaryData();
Object[] o2data = o2.getSummaryData();
- int o1_xt = ((Integer) o1data[idx_upe]) - ((Integer) o1data[idx_ups]);
+ int o1_xt = ((Integer) o1data[idx_upe])
+ - ((Integer) o1data[idx_ups]);
int o1_cat = scoreCategory((String) o1data[idx_mcat]);
- int o2_xt = ((Integer) o2data[idx_upe]-(Integer) o2data[idx_ups]);
+ int o2_xt = ((Integer) o2data[idx_upe] - (Integer) o2data[idx_ups]);
int o2_cat = scoreCategory((String) o2data[idx_mcat]);
- return o2_xt-o1_xt;
+ return o2_xt - o1_xt;
}
});
+ if (filter.equals(
+ ThreeDBStructureChooserQuerySource.FILTER_FIRST_BEST_COVERAGE))
+ {
+ return selected.subList(0, 1);
+ }
return selected;
}