|
| 1 | +package org.biojava.nbio.structure.test.ecod; |
| 2 | + |
| 3 | +import java.io.IOException; |
| 4 | +import java.util.Arrays; |
| 5 | +import java.util.List; |
| 6 | + |
| 7 | +import org.biojava.nbio.structure.Atom; |
| 8 | +import org.biojava.nbio.structure.AtomPositionMap; |
| 9 | +import org.biojava.nbio.structure.Group; |
| 10 | +import org.biojava.nbio.structure.ResidueRange; |
| 11 | +import org.biojava.nbio.structure.ResidueRangeAndLength; |
| 12 | +import org.biojava.nbio.structure.Structure; |
| 13 | +import org.biojava.nbio.structure.StructureException; |
| 14 | +import org.biojava.nbio.structure.StructureTools; |
| 15 | +import org.biojava.nbio.structure.align.util.AtomCache; |
| 16 | +import org.biojava.nbio.structure.ecod.EcodDatabase; |
| 17 | +import org.biojava.nbio.structure.ecod.EcodDomain; |
| 18 | +import org.biojava.nbio.structure.ecod.EcodFactory; |
| 19 | +import org.biojava.nbio.structure.io.LocalPDBDirectory.ObsoleteBehavior; |
| 20 | +import org.slf4j.Logger; |
| 21 | +import org.slf4j.LoggerFactory; |
| 22 | + |
| 23 | +/** |
| 24 | + * This is not a unit test. |
| 25 | + * |
| 26 | + * It is a long-running parsing test, which sequentially parses all ECOD domains. |
| 27 | + * |
| 28 | + * The most common warning is caused by residue ranges with missing terminal CA atoms, |
| 29 | + * which cause a warning to print. |
| 30 | + * |
| 31 | + * develop83 and earlier versions also had a number of invalid ranges, which cause |
| 32 | + * error messages to print. |
| 33 | + * |
| 34 | + * Filtering log4j messages to the 'error' level will filter all but the most |
| 35 | + * grevious errors. |
| 36 | + * |
| 37 | + * @author blivens |
| 38 | + * |
| 39 | + */ |
| 40 | +public class EcodParseTest { |
| 41 | + private static final Logger logger = LoggerFactory.getLogger(EcodParseTest.class); |
| 42 | + |
| 43 | + public static void main(String[] args) throws IOException { |
| 44 | + String ecodVersion = "develop83"; |
| 45 | + EcodDatabase ecod = EcodFactory.getEcodDatabase(ecodVersion); |
| 46 | + AtomCache cache = new AtomCache(); |
| 47 | + cache.setObsoleteBehavior(ObsoleteBehavior.FETCH_OBSOLETE); |
| 48 | + List<EcodDomain> domains = ecod.getAllDomains(); |
| 49 | +// domains = Arrays.asList(ecod.getDomainsById("e1yfbB2")); |
| 50 | +// domains = Arrays.asList(ecod.getDomainsById("e1w50A2")); |
| 51 | + domains = Arrays.asList(ecod.getDomainsById("e2ftlE1")); |
| 52 | + int errors = 0; |
| 53 | + for(EcodDomain d : domains) { |
| 54 | + Atom[] ca1; |
| 55 | + Structure struct; |
| 56 | + try { |
| 57 | + struct = cache.getStructure(d.getPdbId()); |
| 58 | + ca1 = StructureTools.getRepresentativeAtomArray(struct); |
| 59 | + } catch (IOException e) { |
| 60 | + logger.error("Error getting structure for "+d.getDomainId(),e); |
| 61 | + errors++; |
| 62 | + continue; |
| 63 | + } catch (StructureException e) { |
| 64 | + logger.error("Error getting structure for "+d.getDomainId(),e); |
| 65 | + errors++; |
| 66 | + continue; |
| 67 | + } |
| 68 | + |
| 69 | + // Test that the ranges can be parsed |
| 70 | + String rangeStr = d.getRange(); |
| 71 | + AtomPositionMap map = new AtomPositionMap(ca1); |
| 72 | + List<? extends ResidueRange> ranges; |
| 73 | + try { |
| 74 | + // Parses range given in domain |
| 75 | + ranges = ResidueRange.parseMultiple(rangeStr); |
| 76 | + } catch(Exception e) { |
| 77 | + logger.error("Error parsing "+d.getPdbId()+"_"+d.getRange(),e); |
| 78 | + errors++; |
| 79 | + continue; |
| 80 | + } |
| 81 | + boolean clean = true; |
| 82 | + for(ResidueRange r : ranges) { |
| 83 | + if( r == null ) { |
| 84 | + clean = false; |
| 85 | + } |
| 86 | + } |
| 87 | + if( ! clean ) { |
| 88 | + logger.error("Empty range for {}_{}",d.getPdbId(),d.getRange()); |
| 89 | + errors++; |
| 90 | + continue; |
| 91 | + } |
| 92 | + |
| 93 | + |
| 94 | + // Check that the ranges are valid (or at least that they have a group) |
| 95 | + for(ResidueRange range : ranges) { |
| 96 | + try { |
| 97 | + Integer start = map.getPosition(range.getStart()); |
| 98 | + if(start == null) { |
| 99 | + Group g = struct.getChainByPDB(range.getStart().getChainId()).getGroupByPDB(range.getStart()); |
| 100 | + if(g!=null) { |
| 101 | + logger.warn("No CA atom for starting residue "+d.getDomainId()+"_"+range); |
| 102 | + clean = false; |
| 103 | + } else { |
| 104 | + logger.error("Start doesn't exist for "+d.getDomainId()+"_"+range.toString()); |
| 105 | + clean = false; |
| 106 | + } |
| 107 | + } |
| 108 | + } catch(Exception e) { |
| 109 | + logger.error("Start doesn't exist for "+d.getDomainId()+"_"+range.toString(),e); |
| 110 | + clean = false; |
| 111 | + } |
| 112 | + try { |
| 113 | + Integer end = map.getPosition(range.getEnd()); |
| 114 | + if(end == null) { |
| 115 | + Group g = null; |
| 116 | + try { |
| 117 | + g = struct.getChainByPDB(range.getEnd().getChainId()).getGroupByPDB(range.getEnd()); |
| 118 | + } catch(StructureException e ) {} |
| 119 | + if(g!=null) { |
| 120 | + logger.warn("No CA atom for ending residue "+d.getDomainId()+"_"+range); |
| 121 | + clean = false; |
| 122 | + } else { |
| 123 | + logger.error("End doesn't exist for "+d.getDomainId()+"_"+range.toString()); |
| 124 | + clean = false; |
| 125 | + } |
| 126 | + } |
| 127 | + } catch(Exception e) { |
| 128 | + logger.error("End doesn't exist for "+d.getDomainId()+"_"+range.toString(),e); |
| 129 | + clean = false; |
| 130 | + } |
| 131 | + } |
| 132 | + |
| 133 | + // Try to recover from missing residues by trimming them to the residue range |
| 134 | + try { |
| 135 | + // Parses more flexibly, giving only a warning for missing residues |
| 136 | + ranges = ResidueRangeAndLength.parseMultiple(rangeStr,map); |
| 137 | + } catch(Exception e) { |
| 138 | + logger.error("Error parsing "+d.getPdbId()+"_"+d.getRange(),e); |
| 139 | + errors++; |
| 140 | + continue; |
| 141 | + } |
| 142 | + clean = true; |
| 143 | + for(ResidueRange r : ranges) { |
| 144 | + if( r == null ) { |
| 145 | + clean = false; |
| 146 | + } |
| 147 | + } |
| 148 | + if( ! clean ) { |
| 149 | + logger.error("Empty range for {}_{}",d.getPdbId(),d.getRange()); |
| 150 | + errors++; |
| 151 | + continue; |
| 152 | + } |
| 153 | + |
| 154 | + // Test whether we can use it to get a structure |
| 155 | + String pdbRangeStr = String.format("%s.%s",d.getPdbId(),d.getRange()); |
| 156 | + try { |
| 157 | + cache.getStructure(pdbRangeStr); |
| 158 | + } catch(Exception e) { |
| 159 | + logger.error("Can't get range "+pdbRangeStr,e); |
| 160 | + errors++; |
| 161 | + continue; |
| 162 | + } |
| 163 | + |
| 164 | + //All test passed |
| 165 | + logger.info("OK "+d.getDomainId()); |
| 166 | + } |
| 167 | + logger.info("Done. {} errors.",errors); |
| 168 | + } |
| 169 | +} |
0 commit comments