X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fws%2Fdatamodel%2Falphafold%2FPAEContactMatrix.java;h=397a84b2e933e79e42287950c4208c22f28c2fc3;hb=eb3e681d6e82ccdd5d312d1981dfb306e7f479f0;hp=272d21d1f8d766fca9a3d50fb9da324bc05e8f62;hpb=01f8ff469a507604ed961418f7c80f9b245e60ca;p=jalview.git diff --git a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java index 272d21d..397a84b 100644 --- a/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java +++ b/src/jalview/ws/datamodel/alphafold/PAEContactMatrix.java @@ -1,44 +1,52 @@ package jalview.ws.datamodel.alphafold; +import java.io.File; +import java.io.FileInputStream; +import java.io.IOException; import java.util.Iterator; import java.util.List; import java.util.Map; +import org.json.simple.JSONObject; + import jalview.datamodel.ContactListI; import jalview.datamodel.ContactListImpl; import jalview.datamodel.ContactListProviderI; import jalview.datamodel.ContactMatrixI; +import jalview.datamodel.FloatContactMatrix; +import jalview.datamodel.GroupSet; +import jalview.datamodel.SequenceDummy; import jalview.datamodel.SequenceI; +import jalview.io.FileFormatException; +import jalview.util.MapList; import jalview.util.MapUtils; +import jalview.ws.dbsources.EBIAlfaFold; -public class PAEContactMatrix implements ContactMatrixI +/** + * routines and class for holding predicted alignment error matrices as produced + * by alphafold et al. + * + * getContactList(column) returns the vector of predicted alignment errors for + * reference position given by column getElementAt(column, i) returns the + * predicted superposition error for the ith position when column is used as + * reference + * + * Many thanks to Ora Schueler Furman for noticing that earlier development + * versions did not show the PAE oriented correctly + * + * @author jprocter + * + */ +public class PAEContactMatrix extends + MappableContactMatrix implements ContactMatrixI { - SequenceI refSeq = null; - - /** - * the length that refSeq is expected to be (excluding gaps, of course) - */ - int length; - - int maxrow = 0, maxcol = 0; - - int[] indices1, indices2; - - float[][] elements; - - float maxscore; - @SuppressWarnings("unchecked") public PAEContactMatrix(SequenceI _refSeq, Map pae_obj) + throws FileFormatException { - refSeq = _refSeq; - while (refSeq.getDatasetSequence() != null) - { - refSeq = refSeq.getDatasetSequence(); - } + setRefSeq(_refSeq); // convert the lists to primitive arrays and store - length = _refSeq.getEnd() - _refSeq.getStart() + 1; if (!MapUtils.containsAKey(pae_obj, "predicted_aligned_error", "pae")) { @@ -52,32 +60,102 @@ public class PAEContactMatrix implements ContactMatrixI } /** - * parse a sane JSON representation of the pAE + * construct a sequence associated PAE matrix directly from a float array + * + * @param _refSeq + * @param matrix + */ + public PAEContactMatrix(SequenceI _refSeq, float[][] matrix) + { + mappedMatrix = new FloatContactMatrix(matrix); + setRefSeq(_refSeq); + } + + /** + * new matrix with specific mapping to a reference sequence + * + * @param newRefSeq + * @param newFromMapList + * @param elements2 + * @param grps2 + */ + public PAEContactMatrix(SequenceI newRefSeq, MapList newFromMapList, + float[][] elements2, GroupSet grps2) + { + this(newRefSeq, new FloatContactMatrix(elements2, grps2)); + toSeq = newFromMapList; + } + + public PAEContactMatrix(SequenceI _refSeq, + ContactMatrixI floatContactMatrix) + { + mappedMatrix = floatContactMatrix; + setRefSeq(_refSeq); + } + + public PAEContactMatrix(SequenceI _refSeq, MapList newFromMapList, + ContactMatrixI floatContactMatrix) + { + mappedMatrix = floatContactMatrix; + setRefSeq(_refSeq); + toSeq = newFromMapList; + } + + @Override + protected PAEContactMatrix newMappableContactMatrix(SequenceI newRefSeq, + MapList newFromMapList) + { + return new PAEContactMatrix(newRefSeq, newFromMapList, mappedMatrix); + } + + /** + * parse a sane JSON representation of the pAE and update the mappedMatrix * * @param pae_obj */ + @SuppressWarnings("unchecked") private void parse_version_2_pAE(Map pae_obj) { - elements = new float[length][length]; - // this is never going to be reached by the integer rounding.. or is it ? - maxscore = ((Double) MapUtils.getFirst(pae_obj, - "max_predicted_aligned_error", "max_pae")).floatValue(); - Iterator> scoreRows = ((List>) MapUtils - .getFirst(pae_obj, "predicted_aligned_error", "pae")) - .iterator(); + float maxscore = -1; + // look for a maxscore element - if there is one... + try + { + // this is never going to be reached by the integer rounding.. or is it ? + maxscore = ((Double) MapUtils.getFirst(pae_obj, + "max_predicted_aligned_error", "max_pae")).floatValue(); + } catch (Throwable t) + { + // ignore if a key is not found. + } + List> scoreRows = ((List>) MapUtils + .getFirst(pae_obj, "predicted_aligned_error", "pae")); + float[][] elements = new float[scoreRows.size()][scoreRows.size()]; int row = 0, col = 0; - while (scoreRows.hasNext()) + for (List scoreRow : scoreRows) { - Iterator scores = scoreRows.next().iterator(); + Iterator scores = scoreRow.iterator(); while (scores.hasNext()) { - elements[row][col++] = scores.next(); + Object d = scores.next(); + if (d instanceof Double) + { + elements[col][row] = ((Double) d).longValue(); + } + else + { + elements[col][row] = (float) ((Long) d).longValue(); + } + + if (maxscore < elements[col][row]) + { + maxscore = elements[col][row]; + } + col++; } row++; col = 0; } - maxcol = length; - maxrow = length; + mappedMatrix = new FloatContactMatrix(elements); } /** @@ -86,19 +164,18 @@ public class PAEContactMatrix implements ContactMatrixI * * @param pae_obj */ + @SuppressWarnings("unchecked") private void parse_version_1_pAE(Map pae_obj) { // assume indices are with respect to range defined by _refSeq on the // dataset refSeq Iterator rows = ((List) pae_obj.get("residue1")).iterator(); Iterator cols = ((List) pae_obj.get("residue2")).iterator(); - Iterator scores = ((List) pae_obj.get("distance")) - .iterator(); + // two pass - to allocate the elements array - elements = new float[length][length]; - while (scores.hasNext()) + int maxrow = -1, maxcol = -1; + while (rows.hasNext()) { - float escore = scores.next().floatValue(); int row = rows.next().intValue(); int col = cols.next().intValue(); if (maxrow < row) @@ -109,81 +186,97 @@ public class PAEContactMatrix implements ContactMatrixI { maxcol = col; } - elements[row - 1][col - 1] = escore; - } - - maxscore = ((Double) MapUtils.getFirst(pae_obj, - "max_predicted_aligned_error", "max_pae")).floatValue(); - } - @Override - public ContactListI getContactList(final int _column) - { - if (_column < 0 || _column >= elements.length) - { - return null; } - - return new ContactListImpl(new ContactListProviderI() + rows = ((List) pae_obj.get("residue1")).iterator(); + cols = ((List) pae_obj.get("residue2")).iterator(); + Iterator scores = ((List) pae_obj.get("distance")) + .iterator(); + float[][] elements = new float[maxcol][maxrow]; + while (scores.hasNext()) { - @Override - public int getPosition() + float escore = scores.next().floatValue(); + int row = rows.next().intValue(); + int col = cols.next().intValue(); + if (maxrow < row) { - return _column; + maxrow = row; } - - @Override - public int getContactHeight() + if (maxcol < col) { - return maxcol - 1; + maxcol = col; } + elements[col - 1][row - 1] = escore; + } - @Override - public double getContactAt(int column) - { - if (column < 0 || column >= elements[_column].length) - { - return -1; - } - // TODO Auto-generated method stub - return elements[_column][column]; - } - }); + mappedMatrix = new FloatContactMatrix(elements); } @Override - public float getMin() + public String getAnnotDescr() { - return 0; + return "Predicted Alignment Error" + + ((refSeq == null) ? "" : (" for " + refSeq.getName())); } @Override - public float getMax() + public String getAnnotLabel() { - return maxscore; + StringBuilder label = new StringBuilder("PAE Matrix"); + // if (this.getReferenceSeq() != null) + // { + // label.append(":").append(this.getReferenceSeq().getDisplayId(false)); + // } + return label.toString(); } + public static final String PAEMATRIX = "PAE_MATRIX"; + @Override - public boolean hasReferenceSeq() + public String getType() { - return (refSeq != null); + return PAEMATRIX; } - @Override - public SequenceI getReferenceSeq() + public static void validateContactMatrixFile(String fileName) + throws FileFormatException, IOException { - return refSeq; + FileInputStream infile = null; + try + { + infile = new FileInputStream(new File(fileName)); + } catch (Throwable t) + { + new IOException("Couldn't open " + fileName, t); + } + JSONObject paeDict = null; + try + { + paeDict = EBIAlfaFold.parseJSONtoPAEContactMatrix(infile); + } catch (Throwable t) + { + new FileFormatException("Couldn't parse " + fileName + + " as a JSON dict or array containing a dict"); + } + + PAEContactMatrix matrix = new PAEContactMatrix( + new SequenceDummy("Predicted"), (Map) paeDict); + if (matrix.getWidth() <= 0) + { + throw new FileFormatException( + "No data in PAE matrix read from '" + fileName + "'"); + } } @Override - public String getAnnotDescr() + public boolean equals(Object obj) { - return "Predicted Alignment Error for " + refSeq.getName(); + return super.equals(obj); } @Override - public String getAnnotLabel() + public int hashCode() { - return "pAE Matrix"; + return super.hashCode(); } }