Source code

001package org.biojava.nbio.structure.io.cif;
002
003import org.biojava.nbio.structure.*;
004import org.biojava.nbio.structure.xtal.CrystalCell;
005import org.biojava.nbio.structure.xtal.SpaceGroup;
006import org.rcsb.cif.CifBuilder;
007import org.rcsb.cif.model.Category;
008import org.rcsb.cif.model.CifFile;
009import org.rcsb.cif.model.FloatColumnBuilder;
010import org.rcsb.cif.model.IntColumnBuilder;
011import org.rcsb.cif.model.StrColumnBuilder;
012import org.rcsb.cif.schema.StandardSchemata;
013import org.rcsb.cif.schema.mm.MmCifBlockBuilder;
014import org.rcsb.cif.schema.mm.MmCifCategoryBuilder;
015import org.rcsb.cif.schema.mm.MmCifFileBuilder;
016
017import java.util.LinkedHashMap;
018import java.util.List;
019import java.util.Map;
020import java.util.Optional;
021import java.util.function.Consumer;
022import java.util.stream.Collector;
023import java.util.stream.Collectors;
024
025/**
026 * Convert a BioJava object to a CifFile.
027 * @author Sebastian Bittrich
028 * @since 5.3.0
029 */
030public abstract class AbstractCifFileSupplier<S> implements CifFileSupplier<S> {
031    protected CifFile getInternal(Structure structure, List<WrappedAtom> wrappedAtoms) {
032        // for now BioJava only considered 3 categories for create a Cif representation of a structure
033
034        // cell
035        CrystalCell crystalCell = structure.getPDBHeader().getCrystallographicInfo().getCrystalCell();
036        // symmetry
037        SpaceGroup spaceGroup = structure.getPDBHeader().getCrystallographicInfo().getSpaceGroup();
038        // atom_site
039        Category atomSite = wrappedAtoms.stream().collect(toAtomSite());
040        // entity information
041        List<EntityInfo> entityInfos = structure.getEntityInfos();
042
043        MmCifBlockBuilder blockBuilder = CifBuilder.enterFile(StandardSchemata.MMCIF)
044                .enterBlock(structure.getPdbId() == null? "" : structure.getPdbId().getId());
045
046        blockBuilder.enterStructKeywords().enterText()
047        .add(String.join(", ", structure.getPDBHeader().getKeywords()))
048        .leaveColumn().leaveCategory();
049
050        if (atomSite.isDefined() && atomSite.getRowCount() > 0) {
051            // set atom site
052            blockBuilder.addCategory(atomSite);
053        }
054
055        if (crystalCell != null) {
056            // set cell category
057            blockBuilder.enterCell()
058                    .enterLengthA()
059                    .add(crystalCell.getA())
060                    .leaveColumn()
061
062                    .enterLengthB()
063                    .add(crystalCell.getB())
064                    .leaveColumn()
065
066                    .enterLengthC()
067                    .add(crystalCell.getC())
068                    .leaveColumn()
069
070                    .enterAngleAlpha()
071                    .add(crystalCell.getAlpha())
072                    .leaveColumn()
073
074                    .enterAngleBeta()
075                    .add(crystalCell.getBeta())
076                    .leaveColumn()
077
078                    .enterAngleGamma()
079                    .add(crystalCell.getGamma())
080                    .leaveColumn()
081                    .leaveCategory();
082        }
083
084        if (spaceGroup != null) {
085            // set symmetry category
086            blockBuilder.enterSymmetry()
087                    .enterSpaceGroupNameH_M()
088                    .add(spaceGroup.getShortSymbol())
089                    .leaveColumn()
090                    .leaveCategory();
091        }
092
093        if (entityInfos != null) {
094
095            String[] entityIds = new String[entityInfos.size()];
096            String[] entityTypes = new String[entityInfos.size()];
097            String[] entityDescriptions = new String[entityInfos.size()];
098
099            for (int i=0; i<entityInfos.size(); i++) {
100                EntityInfo e = entityInfos.get(i);
101                entityIds[i] = Integer.toString(e.getMolId());
102                entityTypes[i] = e.getType().getEntityType();
103                entityDescriptions[i] = e.getDescription() == null? "?" : e.getDescription();
104            }
105
106            String[] polyEntityIds = entityInfos.stream().filter(e -> e.getType() == EntityType.POLYMER).map(e -> Integer.toString(e.getMolId())).toArray(String[]::new);
107            String[] polyEntitySeqs = entityInfos.stream().filter(e -> e.getType() == EntityType.POLYMER).map(e -> e.getChains().get(0).getSeqResSequence()).toArray(String[]::new);
108
109            blockBuilder.enterEntity()
110                    .enterId()
111                    .add(entityIds)
112                    .leaveColumn()
113
114                    .enterType()
115                    .add(entityTypes)
116                    .leaveColumn()
117
118                    .enterPdbxDescription()
119                    .add(entityDescriptions)
120                    .leaveColumn()
121
122                    .leaveCategory();
123
124            blockBuilder.enterEntityPoly()
125                    .enterEntityId()
126                    .add(polyEntityIds)
127                    .leaveColumn()
128
129                    .enterPdbxSeqOneLetterCodeCan()
130                    .add(polyEntitySeqs)
131                    .leaveColumn()
132
133                    .leaveCategory();
134        }
135
136        return blockBuilder.leaveBlock().leaveFile();
137    }
138
139    protected void handleChain(Chain chain, int model, List<WrappedAtom> wrappedAtoms) {
140        final String chainName = chain.getName();
141        final String chainId = chain.getId();
142        for (Group group : chain.getAtomGroups()) {
143            // The alt locs can have duplicates, since at parsing time we make sure that all alt loc groups have
144            // all atoms (see StructureTools#cleanUpAltLocs)
145            // Thus we have to remove duplicates here by using the atom id
146            // See issue https://github.com/biojava/biojava/issues/778 and
147            // TestAltLocs.testMmcifWritingAllAltlocs/testMmcifWritingPartialAltlocs
148            Map<Integer, WrappedAtom> uniqueAtoms = new LinkedHashMap<>();
149            for (int atomIndex = 0; atomIndex < group.size(); atomIndex++) {
150                Atom atom = group.getAtom(atomIndex);
151                if (atom == null) {
152                    continue;
153                }
154
155                uniqueAtoms.put(atom.getPDBserial(), new WrappedAtom(model, chainName, chainId, atom, atom.getPDBserial()));
156            }
157
158            if (group.hasAltLoc()) {
159                for (Group alt : group.getAltLocs()) {
160                    for (int atomIndex = 0; atomIndex < alt.size(); atomIndex++) {
161                        Atom atom = alt.getAtom(atomIndex);
162                        if (atom == null) {
163                            continue;
164                        }
165
166                        uniqueAtoms.put(atom.getPDBserial(), new WrappedAtom(model, chainName, chainId, atom, atom.getPDBserial()));
167                    }
168                }
169            }
170
171            wrappedAtoms.addAll(uniqueAtoms.values());
172        }
173    }
174
175    /**
176     * Wrapped atoms represent individual atoms enriched with model- and chain-level information. Also, gives control
177     * over the atomId field. Useful to convert structures (and subsets thereof) to their mmCIF representation.
178     */
179    public static class WrappedAtom {
180        private final int model;
181        private final String chainName;
182        private final String chainId;
183        private final Atom atom;
184        private final int atomId;
185
186        /**
187         * Construct a new atoms.
188         * @param model the model number
189         * @param chainName the label_asym_id
190         * @param chainId the auth_asym_id
191         * @param atom the atom instance itself
192         * @param atomId the label_atom_id
193         */
194        public WrappedAtom(int model, String chainName, String chainId, Atom atom, int atomId) {
195            this.model = model;
196            this.chainName = chainName;
197            this.chainId = chainId;
198            this.atom = atom;
199            this.atomId = atomId;
200        }
201
202        public int getModel() {
203            return model;
204        }
205
206        public String getChainName() {
207            return chainName;
208        }
209
210        public String getChainId() {
211            return chainId;
212        }
213
214        public Atom getAtom() {
215            return atom;
216        }
217
218        public int getAtomId() {
219            return atomId;
220        }
221    }
222
223    /**
224     * Collects {@link WrappedAtom} instances into one {@link org.rcsb.cif.schema.mm.AtomSite}.
225     * @return an atom site record containing all atoms
226     */
227    public static Collector<WrappedAtom, ?, Category> toAtomSite() {
228        return Collector.of(AtomSiteCollector::new,
229                AtomSiteCollector::accept,
230                AtomSiteCollector::combine,
231                AtomSiteCollector::get);
232    }
233
234    static class AtomSiteCollector implements Consumer<WrappedAtom> {
235        private final MmCifCategoryBuilder.AtomSiteBuilder atomSiteBuilder;
236        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> groupPDB;
237        private final IntColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> id;
238        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> typeSymbol;
239        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> labelAtomId;
240        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> labelAltId;
241        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> labelCompId;
242        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> labelAsymId;
243        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> labelEntityId;
244        private final IntColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> labelSeqId;
245        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> pdbxPDBInsCode;
246        private final FloatColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> cartnX;
247        private final FloatColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> cartnY;
248        private final FloatColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> cartnZ;
249        private final FloatColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> occupancy;
250        private final FloatColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> bIsoOrEquiv;
251        private final IntColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> authSeqId;
252        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> authCompId;
253        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> authAsymId;
254        private final StrColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> authAtomId;
255        private final IntColumnBuilder<MmCifCategoryBuilder.AtomSiteBuilder, MmCifBlockBuilder, MmCifFileBuilder> pdbxPDBModelNum;
256
257        AtomSiteCollector() {
258            this.atomSiteBuilder = new MmCifCategoryBuilder.AtomSiteBuilder(null);
259            this.groupPDB = atomSiteBuilder.enterGroupPDB();
260            this.id = atomSiteBuilder.enterId();
261            this.typeSymbol = atomSiteBuilder.enterTypeSymbol();
262            this.labelAtomId = atomSiteBuilder.enterLabelAtomId();
263            this.labelAltId = atomSiteBuilder.enterLabelAltId();
264            this.labelCompId = atomSiteBuilder.enterLabelCompId();
265            this.labelAsymId = atomSiteBuilder.enterLabelAsymId();
266            this.labelEntityId = atomSiteBuilder.enterLabelEntityId();
267            this.labelSeqId = atomSiteBuilder.enterLabelSeqId();
268            this.pdbxPDBInsCode = atomSiteBuilder.enterPdbxPDBInsCode();
269            this.cartnX = atomSiteBuilder.enterCartnX();
270            this.cartnY = atomSiteBuilder.enterCartnY();
271            this.cartnZ = atomSiteBuilder.enterCartnZ();
272            this.occupancy = atomSiteBuilder.enterOccupancy();
273            this.bIsoOrEquiv = atomSiteBuilder.enterBIsoOrEquiv();
274            this.authSeqId = atomSiteBuilder.enterAuthSeqId();
275            this.authCompId = atomSiteBuilder.enterAuthCompId();
276            this.authAsymId = atomSiteBuilder.enterAuthAsymId();
277            this.authAtomId = atomSiteBuilder.enterAuthAtomId();
278            this.pdbxPDBModelNum = atomSiteBuilder.enterPdbxPDBModelNum();
279        }
280
281        @Override
282        public void accept(WrappedAtom wrappedAtom) {
283            Atom atom = wrappedAtom.getAtom();
284            Group group = atom.getGroup();
285            Chain chain = group.getChain();
286
287            groupPDB.add(group.getType().equals(GroupType.HETATM) ? "HETATM" : "ATOM");
288            id.add(wrappedAtom.getAtomId());
289            Element element = atom.getElement();
290            typeSymbol.add(element.equals(Element.R) ? "X" : element.toString().toUpperCase());
291            labelAtomId.add(atom.getName());
292            Character altLoc = atom.getAltLoc();
293            if (altLoc == null || altLoc == ' ') {
294                labelAltId.markNextNotPresent();
295            } else {
296                labelAltId.add(String.valueOf(altLoc));
297            }
298            labelCompId.add(group.getPDBName());
299            labelAsymId.add(wrappedAtom.getChainId());
300            String entityId = "0";
301            int seqId = group.getResidueNumber().getSeqNum();
302            if (chain.getEntityInfo() != null) {
303                entityId = Integer.toString(chain.getEntityInfo().getMolId());
304                if (chain.getEntityInfo().getType() == EntityType.POLYMER) {
305                    // this only makes sense for polymeric chains, non-polymer chains will never have seqres groups and
306                    // there's no point in calling getAlignedResIndex
307                    seqId = chain.getEntityInfo().getAlignedResIndex(group, chain);
308                }
309            }
310            labelEntityId.add(entityId);
311            labelSeqId.add(seqId);
312            String insCode = "";
313            if (group.getResidueNumber().getInsCode() != null) {
314                insCode = Character.toString(group.getResidueNumber().getInsCode());
315            }
316            if (insCode.isEmpty()) {
317                pdbxPDBInsCode.markNextUnknown();
318            } else {
319                pdbxPDBInsCode.add(insCode);
320            }
321            cartnX.add(atom.getX());
322            cartnY.add(atom.getY());
323            cartnZ.add(atom.getZ());
324            occupancy.add(atom.getOccupancy());
325            bIsoOrEquiv.add(atom.getTempFactor());
326            authSeqId.add(group.getResidueNumber().getSeqNum());
327            authCompId.add(group.getPDBName());
328            authAsymId.add(wrappedAtom.getChainName());
329            authAtomId.add(atom.getName());
330            pdbxPDBModelNum.add(wrappedAtom.getModel());
331        }
332
333        AtomSiteCollector combine(AtomSiteCollector other) {
334            throw new UnsupportedOperationException("impl by calling addAll for all collections");
335        }
336
337        Category get() {
338            groupPDB.leaveColumn();
339            id.leaveColumn();
340            typeSymbol.leaveColumn();
341            labelAtomId.leaveColumn();
342            labelAltId.leaveColumn();
343            labelCompId.leaveColumn();
344            labelAsymId.leaveColumn();
345            labelEntityId.leaveColumn();
346            labelSeqId.leaveColumn();
347            pdbxPDBInsCode.leaveColumn();
348            cartnX.leaveColumn();
349            cartnY.leaveColumn();
350            cartnZ.leaveColumn();
351            occupancy.leaveColumn();
352            bIsoOrEquiv.leaveColumn();
353            authSeqId.leaveColumn();
354            authCompId.leaveColumn();
355            authAsymId.leaveColumn();
356            authAtomId.leaveColumn();
357            pdbxPDBModelNum.leaveColumn();
358            return atomSiteBuilder.build();
359        }
360    }
361}