1. Project Clover database Fri Dec 6 2024 13:47:14 GMT
  2. Package jalview.ws.dbsources

File EBIAlfaFold.java

 

Coverage histogram

../../../img/srcFileCovDistChart3.png
52% of files have more coverage

Code metrics

74
170
24
1
678
467
82
0.48
7.08
24
3.42

Classes

Class
Line #
Actions
EBIAlfaFold 69 170 82
0.302238830.2%
 

Contributing tests

This file is covered by 32 tests. .

Source view

1   
2    /*
3    * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
4    * Copyright (C) $$Year-Rel$$ The Jalview Authors
5    *
6    * This file is part of Jalview.
7    *
8    * Jalview is free software: you can redistribute it and/or
9    * modify it under the terms of the GNU General Public License
10    * as published by the Free Software Foundation, either version 3
11    * of the License, or (at your option) any later version.
12    *
13    * Jalview is distributed in the hope that it will be useful, but
14    * WITHOUT ANY WARRANTY; without even the implied warranty
15    * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
16    * PURPOSE. See the GNU General Public License for more details.
17    *
18    * You should have received a copy of the GNU General Public License
19    * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
20    * The Jalview Authors are detailed in the 'AUTHORS' file.
21    */
22    package jalview.ws.dbsources;
23   
24    import java.io.File;
25    import java.io.FileInputStream;
26    import java.io.FileNotFoundException;
27    import java.io.IOException;
28    import java.io.InputStream;
29    import java.util.ArrayList;
30    import java.util.Date;
31    import java.util.HashMap;
32    import java.util.List;
33    import java.util.Map;
34   
35    import org.json.simple.JSONArray;
36    import org.json.simple.JSONObject;
37    import org.json.simple.parser.ParseException;
38   
39    import com.stevesoft.pat.Regex;
40   
41    import jalview.api.FeatureSettingsModelI;
42    import jalview.bin.Console;
43    import jalview.datamodel.AlignmentAnnotation;
44    import jalview.datamodel.AlignmentI;
45    import jalview.datamodel.ContactMatrixI;
46    import jalview.datamodel.DBRefEntry;
47    import jalview.datamodel.GroupSet;
48    import jalview.datamodel.PDBEntry;
49    import jalview.datamodel.SequenceFeature;
50    import jalview.datamodel.SequenceI;
51    import jalview.gui.Desktop;
52    import jalview.io.DataSourceType;
53    import jalview.io.FileFormat;
54    import jalview.io.FileFormatI;
55    import jalview.io.FormatAdapter;
56    import jalview.io.PDBFeatureSettings;
57    import jalview.structure.StructureImportSettings.TFType;
58    import jalview.structure.StructureMapping;
59    import jalview.structure.StructureSelectionManager;
60    import jalview.util.MessageManager;
61    import jalview.util.Platform;
62    import jalview.ws.datamodel.alphafold.PAEContactMatrix;
63    import jalview.ws.utils.UrlDownloadClient;
64   
65    /**
66    * @author JimP
67    *
68    */
 
69    public class EBIAlfaFold extends EbiFileRetrievedProxy
70    {
71    private static final String SEPARATOR = "|";
72   
73    private static final String COLON = ":";
74   
75    private static final int PDB_ID_LENGTH = 4;
76   
77    private static String AF_VERSION = "3";
78   
 
79  0 toggle public EBIAlfaFold()
80    {
81  0 super();
82    }
83   
84    /*
85    * (non-Javadoc)
86    *
87    * @see jalview.ws.DbSourceProxy#getAccessionSeparator()
88    */
 
89  0 toggle @Override
90    public String getAccessionSeparator()
91    {
92  0 return null;
93    }
94   
95    /*
96    * (non-Javadoc)
97    *
98    * @see jalview.ws.DbSourceProxy#getAccessionValidator()
99    */
 
100  0 toggle @Override
101    public Regex getAccessionValidator()
102    {
103  0 Regex validator = new Regex("(AF-[A-Z]+[0-9]+[A-Z0-9]+-F1)");
104  0 validator.setIgnoreCase(true);
105  0 return validator;
106    }
107   
108    /*
109    * (non-Javadoc)
110    *
111    * @see jalview.ws.DbSourceProxy#getDbSource()
112    */
 
113  0 toggle @Override
114    public String getDbSource()
115    {
116  0 return "ALPHAFOLD";
117    }
118   
119    /*
120    * (non-Javadoc)
121    *
122    * @see jalview.ws.DbSourceProxy#getDbVersion()
123    */
 
124  0 toggle @Override
125    public String getDbVersion()
126    {
127  0 return "1";
128    }
129   
 
130  0 toggle public static String getAlphaFoldCifDownloadUrl(String id, String vnum)
131    {
132  0 if (vnum == null || vnum.length() == 0)
133    {
134  0 vnum = AF_VERSION;
135    }
136  0 return "https://alphafold.ebi.ac.uk/files/" + id + "-model_v" + vnum
137    + ".cif";
138    }
139   
 
140  6 toggle public static String getAlphaFoldPaeDownloadUrl(String id, String vnum)
141    {
142  6 if (vnum == null || vnum.length() == 0)
143    {
144  0 vnum = AF_VERSION;
145    }
146  6 return "https://alphafold.ebi.ac.uk/files/" + id
147    + "-predicted_aligned_error_v" + vnum + ".json";
148    }
149   
150    /*
151    * (non-Javadoc)
152    *
153    * @see jalview.ws.DbSourceProxy#getSequenceRecords(java.lang.String[])
154    */
 
155  0 toggle @Override
156    public AlignmentI getSequenceRecords(String queries) throws Exception
157    {
158  0 return getSequenceRecords(queries, null);
159    }
160   
 
161  0 toggle public AlignmentI getSequenceRecords(String queries, String retrievalUrl)
162    throws Exception
163    {
164  0 AlignmentI pdbAlignment = null;
165  0 String chain = null;
166  0 String id = null;
167  0 if (queries.indexOf(COLON) > -1)
168    {
169  0 chain = queries.substring(queries.indexOf(COLON) + 1);
170  0 id = queries.substring(0, queries.indexOf(COLON));
171    }
172    else
173    {
174  0 id = queries;
175    }
176   
177  0 if (!isValidReference(id))
178    {
179  0 jalview.bin.Console.errPrintln(
180    "(AFClient) Ignoring invalid alphafold query: '" + id + "'");
181  0 stopQuery();
182  0 return null;
183    }
184  0 String alphaFoldCif = getAlphaFoldCifDownloadUrl(id, AF_VERSION);
185  0 if (retrievalUrl != null)
186    {
187  0 alphaFoldCif = retrievalUrl;
188    }
189   
190  0 try
191    {
192  0 File tmpFile = File.createTempFile(id, ".cif");
193  0 Console.debug("Retrieving structure file for " + id + " from "
194    + alphaFoldCif);
195  0 UrlDownloadClient.download(alphaFoldCif, tmpFile);
196   
197    // may not need this check ?
198  0 file = tmpFile.getAbsolutePath();
199  0 if (file == null)
200    {
201  0 return null;
202    }
203    // TODO Get the PAE file somewhere around here and remove from JmolParser
204   
205  0 pdbAlignment = importDownloadedStructureFromUrl(alphaFoldCif, tmpFile,
206    id, chain, getDbSource(), getDbVersion());
207   
208  0 if (pdbAlignment == null || pdbAlignment.getHeight() < 1)
209    {
210  0 throw new Exception(MessageManager.formatMessage(
211    "exception.no_pdb_records_for_chain", new String[]
212  0 { id, ((chain == null) ? "' '" : chain) }));
213    }
214    // done during structure retrieval
215    // retrieve_AlphaFold_pAE(id, pdbAlignment, retrievalUrl);
216   
217    } catch (Exception ex) // Problem parsing PDB file
218    {
219  0 stopQuery();
220  0 throw (ex);
221    }
222  0 return pdbAlignment;
223    }
224   
225    /**
226    * get an alphafold pAE for the given id and return the File object of the
227    * downloaded (temp) file
228    *
229    * @param id
230    * @param pdbAlignment
231    * @param retrievalUrl
232    * - URL of .mmcif from EBI-AlphaFold - will be used to generate the
233    * pAE URL automatically
234    * @throws IOException
235    * @throws Exception
236    */
 
237  6 toggle public static File fetchAlphaFoldPAE(String id, String retrievalUrl)
238    throws IOException
239    {
240    // import PAE as contact matrix - assume this will work if there was a
241    // model
242  6 String paeURL = getAlphaFoldPaeDownloadUrl(id, AF_VERSION);
243   
244  6 if (retrievalUrl != null)
245    {
246    // manufacture the PAE url from a url like ...-model-vN.cif
247  0 paeURL = retrievalUrl.replace("model", "predicted_aligned_error")
248    .replace(".cif", ".json");
249    }
250   
251    // check the cache
252  6 File pae = paeDownloadCache.get(paeURL);
253  6 if (pae != null && pae.exists() && (new Date().getTime()
254    - pae.lastModified()) < PAE_CACHE_STALE_TIME)
255    {
256  5 Console.debug(
257    "Using existing file in PAE cache for '" + paeURL + "'");
258  5 return pae;
259    }
260   
261  1 try
262    {
263  1 pae = File.createTempFile(id == null ? "af_pae" : id, "pae_json");
264    } catch (IOException e)
265    {
266  0 e.printStackTrace();
267    }
268  1 Console.debug("Downloading pae from " + paeURL + " to " + pae.toString()
269    + "");
270  1 try
271    {
272  1 UrlDownloadClient.download(paeURL, pae);
273    } catch (IOException e)
274    {
275  0 throw e;
276    }
277    // cache and it if successful
278  1 paeDownloadCache.put(paeURL, pae);
279  1 return pae;
280    }
281   
282    /**
283    * get an alphafold pAE for the given id, and add it to sequence 0 in
284    * pdbAlignment (assuming it came from structurefile parser).
285    *
286    * @param id
287    * @param pdbAlignment
288    * @param retrievalUrl
289    * - URL of .mmcif from EBI-AlphaFold - will be used to generate the
290    * pAE URL automatically
291    * @throws IOException
292    * @throws Exception
293    */
 
294  0 toggle public static void retrieve_AlphaFold_pAE(String id,
295    AlignmentI pdbAlignment, String retrievalUrl) throws IOException
296    {
297  0 File pae = fetchAlphaFoldPAE(id, retrievalUrl);
298  0 addAlphaFoldPAE(pdbAlignment, pae, 0, null, false, false, null);
299    }
300   
 
301  42 toggle public static void addAlphaFoldPAE(AlignmentI pdbAlignment, File pae,
302    int index, String id, boolean isStruct, boolean isStructId,
303    String label)
304    {
305  42 FileInputStream paeInput = null;
306  42 try
307    {
308  42 paeInput = new FileInputStream(pae);
309    } catch (FileNotFoundException e)
310    {
311  0 Console.error(
312    "Could not find pAE file '" + pae.getAbsolutePath() + "'", e);
313  0 return;
314    }
315   
316  42 if (isStruct)
317    {
318    // ###### WRITE A TEST for this bit of the logic addAlphaFoldPAE with
319    // different params.
320  0 StructureSelectionManager ssm = StructureSelectionManager
321    .getStructureSelectionManager(Desktop.instance);
322  0 if (ssm != null)
323    {
324  0 String structFilename = isStructId ? ssm.findFileForPDBId(id) : id;
325  0 addPAEToStructure(ssm, structFilename, pae, label);
326    }
327   
328    }
329    else
330    {
331    // attach to sequence?!
332  42 try
333    {
334  42 if (!importPaeJSONAsContactMatrixToSequence(pdbAlignment, paeInput,
335    index, id, label))
336    {
337  0 Console.warn("Could not import contact matrix from '"
338    + pae.getAbsolutePath() + "' to sequence.");
339    }
340    } catch (IOException e1)
341    {
342  0 Console.error("Error when importing pAE file '"
343    + pae.getAbsolutePath() + "'", e1);
344    } catch (ParseException e2)
345    {
346  0 Console.error("Error when parsing pAE file '"
347    + pae.getAbsolutePath() + "'", e2);
348    }
349    }
350   
351    }
352   
 
353  0 toggle public static void addPAEToStructure(StructureSelectionManager ssm,
354    String structFilename, File pae, String label)
355    {
356  0 FileInputStream paeInput = null;
357  0 try
358    {
359  0 paeInput = new FileInputStream(pae);
360    } catch (FileNotFoundException e)
361    {
362  0 Console.error(
363    "Could not find pAE file '" + pae.getAbsolutePath() + "'", e);
364  0 return;
365    }
366  0 if (ssm == null)
367    {
368  0 ssm = StructureSelectionManager
369    .getStructureSelectionManager(Desktop.instance);
370    }
371  0 if (ssm != null)
372    {
373  0 StructureMapping[] smArray = ssm.getMapping(structFilename);
374   
375  0 try
376    {
377  0 if (!importPaeJSONAsContactMatrixToStructure(smArray, paeInput,
378    label))
379    {
380  0 Console.warn("Could not import contact matrix from '"
381    + pae.getAbsolutePath() + "' to structure.");
382    }
383    } catch (IOException e1)
384    {
385  0 Console.error("Error when importing pAE file '"
386    + pae.getAbsolutePath() + "'", e1);
387    } catch (ParseException e2)
388    {
389  0 Console.error("Error when parsing pAE file '"
390    + pae.getAbsolutePath() + "'", e2);
391    }
392    }
393    }
394   
395    /**
396    * parses the given pAE matrix and adds it to sequence 0 in the given
397    * alignment
398    *
399    * @param pdbAlignment
400    * @param pae_input
401    * @return true if there was a pAE matrix added
402    * @throws ParseException
403    * @throws IOException
404    * @throws Exception
405    */
 
406  42 toggle public static boolean importPaeJSONAsContactMatrixToSequence(
407    AlignmentI pdbAlignment, InputStream pae_input, int index,
408    String seqId, String label) throws IOException, ParseException
409    {
410  42 SequenceI sequence = null;
411  42 if (seqId == null)
412    {
413  42 int seqToGet = index > 0 ? index : 0;
414  42 sequence = pdbAlignment.getSequenceAt(seqToGet);
415    }
416  42 if (sequence == null)
417    {
418  0 SequenceI[] sequences = pdbAlignment.findSequenceMatch(seqId);
419  0 if (sequences == null || sequences.length < 1)
420    {
421  0 Console.warn("Could not find sequence with id '" + seqId
422    + "' to attach pAE matrix to. Ignoring matrix.");
423  0 return false;
424    }
425    else
426    {
427  0 sequence = sequences[0]; // just use the first sequence with this seqId
428    }
429    }
430  42 if (sequence == null)
431    {
432  0 return false;
433    }
434  42 return importPaeJSONAsContactMatrixToSequence(pdbAlignment, pae_input,
435    sequence, label);
436    }
437   
 
438  42 toggle public static boolean importPaeJSONAsContactMatrixToSequence(
439    AlignmentI pdbAlignment, InputStream pae_input,
440    SequenceI sequence, String label)
441    throws IOException, ParseException
442    {
443  42 JSONObject paeDict = parseJSONtoPAEContactMatrix(pae_input);
444  42 if (paeDict == null)
445    {
446  0 Console.debug("JSON file did not parse properly.");
447  0 return false;
448    }
449  42 ContactMatrixI matrix = new PAEContactMatrix(sequence,
450    (Map<String, Object>) paeDict);
451   
452  42 AlignmentAnnotation cmannot = sequence.addContactList(matrix);
453  42 if (label != null)
454  0 cmannot.label = label;
455  42 pdbAlignment.addAnnotation(cmannot);
456   
457  42 return true;
458    }
459   
 
460  48 toggle public static JSONObject parseJSONtoPAEContactMatrix(
461    InputStream pae_input) throws IOException, ParseException
462    {
463  48 Object paeJson = Platform.parseJSON(pae_input);
464  48 JSONObject paeDict = null;
465  48 if (paeJson instanceof JSONObject)
466    {
467  15 paeDict = (JSONObject) paeJson;
468    }
469  33 else if (paeJson instanceof JSONArray)
470    {
471  33 JSONArray jsonArray = (JSONArray) paeJson;
472  33 if (jsonArray.size() > 0)
473  33 paeDict = (JSONObject) jsonArray.get(0);
474    }
475   
476  48 return paeDict;
477    }
478   
479    // ###### TEST THIS
 
480  3 toggle public static boolean importPaeJSONAsContactMatrixToStructure(
481    StructureMapping[] smArray, InputStream paeInput, String label)
482    throws IOException, ParseException
483    {
484  3 boolean someDone = false;
485  3 for (StructureMapping sm : smArray)
486    {
487  3 boolean thisDone = importPaeJSONAsContactMatrixToStructure(sm,
488    paeInput, label);
489  3 someDone |= thisDone;
490    }
491  3 return someDone;
492    }
493   
 
494  3 toggle public static boolean importPaeJSONAsContactMatrixToStructure(
495    StructureMapping sm, InputStream paeInput, String label)
496    throws IOException, ParseException
497    {
498  3 JSONObject pae_obj = parseJSONtoPAEContactMatrix(paeInput);
499  3 if (pae_obj == null)
500    {
501  0 Console.debug("JSON file did not parse properly.");
502  0 return false;
503    }
504   
505  3 SequenceI seq = sm.getSequence();
506  3 ContactMatrixI matrix = new PAEContactMatrix(seq,
507    (Map<String, Object>) pae_obj);
508  3 AlignmentAnnotation cmannot = sm.getSequence().addContactList(matrix);
509    /* this already happens in Sequence.addContactList()
510    seq.addAlignmentAnnotation(cmannot);
511    */
512  3 return true;
513    }
514   
515    /**
516    * general purpose structure importer - designed to yield alignment useful for
517    * transfer of annotation to associated sequences
518    *
519    * @param alphaFoldCif
520    * @param tmpFile
521    * @param id
522    * @param chain
523    * @param dbSource
524    * @param dbVersion
525    * @return
526    * @throws Exception
527    */
 
528  0 toggle public static AlignmentI importDownloadedStructureFromUrl(
529    String alphaFoldCif, File tmpFile, String id, String chain,
530    String dbSource, String dbVersion) throws Exception
531    {
532  0 String file = tmpFile.getAbsolutePath();
533    // todo get rid of Type and use FileFormatI instead?
534  0 FileFormatI fileFormat = FileFormat.MMCif;
535  0 TFType tempfacType = TFType.PLDDT;
536  0 AlignmentI pdbAlignment = new FormatAdapter().readFile(tmpFile, file,
537    DataSourceType.FILE, fileFormat, tempfacType);
538   
539  0 if (pdbAlignment != null)
540    {
541  0 List<SequenceI> toremove = new ArrayList<SequenceI>();
542  0 for (SequenceI pdbcs : pdbAlignment.getSequences())
543    {
544  0 String chid = null;
545    // Mapping map=null;
546  0 for (PDBEntry pid : pdbcs.getAllPDBEntries())
547    {
548  0 if (pid.getFile() == file)
549    {
550  0 chid = pid.getChainCode();
551    }
552    }
553  0 if (chain == null || (chid != null && (chid.equals(chain)
554    || chid.trim().equals(chain.trim())
555    || (chain.trim().length() == 0 && chid.equals("_")))))
556    {
557    // FIXME seems to result in 'PDB|1QIP|1qip|A' - 1QIP is redundant.
558    // TODO: suggest simplify naming to 1qip|A as default name defined
559  0 pdbcs.setName(id + SEPARATOR + pdbcs.getName());
560    // Might need to add more metadata to the PDBEntry object
561    // like below
562    /*
563    * PDBEntry entry = new PDBEntry(); // Construct the PDBEntry
564    * entry.setId(id); if (entry.getProperty() == null)
565    * entry.setProperty(new Hashtable());
566    * entry.getProperty().put("chains", pdbchain.id + "=" +
567    * sq.getStart() + "-" + sq.getEnd());
568    * sq.getDatasetSequence().addPDBId(entry);
569    */
570    // Add PDB DB Refs
571    // We make a DBRefEtntry because we have obtained the PDB file from
572    // a
573    // verifiable source
574    // JBPNote - PDB DBRefEntry should also carry the chain and mapping
575    // information
576  0 if (dbSource != null)
577    {
578  0 DBRefEntry dbentry = new DBRefEntry(dbSource,
579   
580  0 dbVersion, (chid == null ? id : id + chid));
581    // dbentry.setMap()
582  0 pdbcs.addDBRef(dbentry);
583    // update any feature groups
584  0 List<SequenceFeature> allsf = pdbcs.getFeatures()
585    .getAllFeatures();
586  0 List<SequenceFeature> newsf = new ArrayList<SequenceFeature>();
587  0 if (allsf != null && allsf.size() > 0)
588    {
589  0 for (SequenceFeature f : allsf)
590    {
591  0 if (file.equals(f.getFeatureGroup()))
592    {
593  0 f = new SequenceFeature(f, f.type, f.begin, f.end, id,
594    f.score);
595    }
596  0 newsf.add(f);
597    }
598  0 pdbcs.setSequenceFeatures(newsf);
599    }
600    }
601    }
602    else
603    {
604    // mark this sequence to be removed from the alignment
605    // - since it's not from the right chain
606  0 toremove.add(pdbcs);
607    }
608    }
609    // now remove marked sequences
610  0 for (SequenceI pdbcs : toremove)
611    {
612  0 pdbAlignment.deleteSequence(pdbcs);
613  0 if (pdbcs.getAnnotation() != null)
614    {
615  0 for (AlignmentAnnotation aa : pdbcs.getAnnotation())
616    {
617  0 pdbAlignment.deleteAnnotation(aa);
618    }
619    }
620    }
621    }
622  0 return pdbAlignment;
623    }
624   
625    /*
626    * (non-Javadoc)
627    *
628    * @see jalview.ws.DbSourceProxy#isValidReference(java.lang.String)
629    */
 
630  0 toggle @Override
631    public boolean isValidReference(String accession)
632    {
633  0 Regex r = getAccessionValidator();
634  0 return r.search(accession.trim());
635    }
636   
637    /**
638    * human glyoxalase
639    */
 
640  0 toggle @Override
641    public String getTestQuery()
642    {
643  0 return "AF-O15552-F1";
644    }
645   
 
646  0 toggle @Override
647    public String getDbName()
648    {
649  0 return "ALPHAFOLD"; // getDbSource();
650    }
651   
 
652  0 toggle @Override
653    public int getTier()
654    {
655  0 return 0;
656    }
657   
658    /**
659    * Returns a descriptor for suitable feature display settings with
660    * <ul>
661    * <li>ResNums or insertions features visible</li>
662    * <li>insertions features coloured red</li>
663    * <li>ResNum features coloured by label</li>
664    * <li>Insertions displayed above (on top of) ResNums</li>
665    * </ul>
666    */
 
667  0 toggle @Override
668    public FeatureSettingsModelI getFeatureColourScheme()
669    {
670  0 return new PDBFeatureSettings();
671    }
672   
673    // days * 86400000
674    private static final long PAE_CACHE_STALE_TIME = 1 * 86400000;
675   
676    private static Map<String, File> paeDownloadCache = new HashMap<>();
677   
678    }