src/jalview/ws/dbsources/PDBRestClient.java

   1 package jalview.ws.dbsources;
   2
   3 import jalview.ws.uimodel.PDBRestRequest;
   4 import jalview.ws.uimodel.PDBRestResponse;
   5 import jalview.ws.uimodel.PDBRestResponse.PDBResponseSummary;
   6
   7 import java.util.ArrayList;
   8 import java.util.Collection;
   9 import java.util.Iterator;
  10 import java.util.List;
  11
  12 import javax.ws.rs.core.MediaType;
  13
  14 import org.json.simple.JSONArray;
  15 import org.json.simple.JSONObject;
  16 import org.json.simple.parser.JSONParser;
  17 import org.json.simple.parser.ParseException;
  18
  19 import com.sun.jersey.api.client.Client;
  20 import com.sun.jersey.api.client.ClientResponse;
  21 import com.sun.jersey.api.client.WebResource;
  22 import com.sun.jersey.api.client.config.ClientConfig;
  23 import com.sun.jersey.api.client.config.DefaultClientConfig;
  24 import com.sun.jersey.api.json.JSONConfiguration;
  25
  26 public class PDBRestClient
  27 {
  28   private String pdbSearchEndpoint = "http://wwwdev.ebi.ac.uk/pdbe/search/pdb/select?";
  29
  30   /**
  31    * Takes a PDBRestRequest object and returns a response upon execution
  32    *
  33    * @param pdbRestRequest
  34    *          the pdbRequest to be sent
  35    * @return the pdbResponse object for the given pdbRequest
  36    */
  37   public PDBRestResponse executeRequest(PDBRestRequest pdbRestRequest)
  38   {
  39     ClientConfig clientConfig = new DefaultClientConfig();
  40     clientConfig.getFeatures().put(JSONConfiguration.FEATURE_POJO_MAPPING,
  41             Boolean.TRUE);
  42     Client client = Client.create(clientConfig);
  43
  44     String query = pdbRestRequest.getFieldToSearchBy()
  45             + pdbRestRequest.getSearchTerm()
  46             + ((pdbRestRequest.isAllowEmptySeq()) ? ""
  47                     : " AND molecule_sequence:['' TO *]");
  48
  49     String wantedFields = getPDBDocFieldsAsCommaDelimitedString(pdbRestRequest
  50             .getWantedFields());
  51
  52     String responseSize = (pdbRestRequest.getResponseSize() == 0) ? "200"
  53             : String.valueOf(pdbRestRequest.getResponseSize());
  54     String sortParam = (pdbRestRequest.getFieldToSortBy() == null || pdbRestRequest
  55             .getFieldToSortBy().trim().isEmpty()) ? ""
  56  : (pdbRestRequest
  57             .getFieldToSortBy() + (pdbRestRequest.isAscending() ? " asc"
  58             : " desc"));
  59
  60     WebResource webResource = client.resource(pdbSearchEndpoint)
  61             .queryParam("wt", "json").queryParam("fl", wantedFields)
  62             .queryParam("rows", responseSize)
  63             .queryParam("q", query)
  64             .queryParam("sort", sortParam);
  65     ClientResponse clientResponse = webResource.accept(
  66             MediaType.APPLICATION_JSON).get(ClientResponse.class);
  67
  68     String responseString = clientResponse.getEntity(String.class);
  69     if (clientResponse.getStatus() != 200)
  70     {
  71       if (clientResponse.getStatus() == 400)
  72       {
  73         throw new RuntimeException(parseJsonExceptionString(responseString));
  74       }
  75       else
  76       {
  77       throw new RuntimeException("Failed : HTTP error code : "
  78               + clientResponse.getStatus());
  79       }
  80     }
  81     clientResponse = null;
  82     client = null;
  83     return parsePDBJsonResponse(responseString, pdbRestRequest);
  84   }
  85
  86   /**
  87    * Process error response from PDB server if/when one occurs.
  88    *
  89    * @param jsonResponse
  90    *          the json string containing error message from the server
  91    * @return the processed error message from the json string
  92    */
  93   public static String parseJsonExceptionString(String jsonErrorResponse)
  94   {
  95     String errorMessage = "RunTime error";
  96     try
  97     {
  98       JSONParser jsonParser = new JSONParser();
  99       JSONObject jsonObj = (JSONObject) jsonParser.parse(jsonErrorResponse);
 100       JSONObject errorResponse = (JSONObject) jsonObj.get("error");
 101       errorMessage = errorResponse.get("msg").toString();
 102
 103       JSONObject responseHeader = (JSONObject) jsonObj
 104               .get("responseHeader");
 105       errorMessage += responseHeader.get("params").toString();
 106     } catch (ParseException e)
 107     {
 108       e.printStackTrace();
 109     }
 110     return errorMessage;
 111   }
 112
 113   /**
 114    * Parses json response string from PDB REST API to a PDBRestResponse
 115    * instance. The parsed response is dynamic and based upon some of the request
 116    * parameters.
 117    *
 118    * @param pdbJsonResponseString
 119    *          the json string to be parsed
 120    * @param pdbRestRequest
 121    *          the request object which contains parameters used to process the
 122    *          json string
 123    * @return
 124    */
 125   @SuppressWarnings("unchecked")
 126   public static PDBRestResponse parsePDBJsonResponse(
 127           String pdbJsonResponseString,
 128           PDBRestRequest pdbRestRequest)
 129   {
 130     PDBRestResponse searchResult = new PDBRestResponse();
 131     List<PDBResponseSummary> result = null;
 132     try
 133     {
 134       JSONParser jsonParser = new JSONParser();
 135       JSONObject jsonObj = (JSONObject) jsonParser
 136               .parse(pdbJsonResponseString);
 137
 138       JSONObject pdbResponse = (JSONObject) jsonObj.get("response");
 139       String queryTime = ((JSONObject) jsonObj.get("responseHeader")).get(
 140               "QTime").toString();
 141       int numFound = Integer
 142               .valueOf(pdbResponse.get("numFound").toString());
 143       if (numFound > 0)
 144       {
 145         result = new ArrayList<PDBResponseSummary>();
 146         JSONArray docs = (JSONArray) pdbResponse.get("docs");
 147         for (Iterator<JSONObject> docIter = docs.iterator(); docIter
 148                 .hasNext();)
 149         {
 150           JSONObject doc = docIter.next();
 151           result.add(searchResult.new PDBResponseSummary(doc, pdbRestRequest));
 152         }
 153         searchResult.setNumberOfItemsFound(numFound);
 154         searchResult.setResponseTime(queryTime);
 155         searchResult.setSearchSummary(result);
 156       }
 157     } catch (ParseException e)
 158     {
 159       e.printStackTrace();
 160     }
 161     return searchResult;
 162   }
 163
 164   /**
 165    * Takes a collection of PDBDocField and converts its code values into a comma
 166    * delimited string.
 167    *
 168    * @param pdbDocfields
 169    * @return
 170    */
 171   public static String getPDBDocFieldsAsCommaDelimitedString(
 172           Collection<PDBDocField> pdbDocfields)
 173   {
 174     String result = "";
 175     if (pdbDocfields != null && !pdbDocfields.isEmpty())
 176     {
 177       StringBuilder returnedFields = new StringBuilder();
 178       for (PDBDocField field : pdbDocfields)
 179       {
 180         returnedFields.append(",").append(field.getCode());
 181       }
 182       returnedFields.deleteCharAt(0);
 183       result = returnedFields.toString();
 184     }
 185     return result;
 186   }
 187
 188   /**
 189    * Determines the column index for 'PDB Id' Fields in the dynamic summary
 190    * table. The PDB Id serves as a unique identifier for a given row in the
 191    * summary table
 192    *
 193    * @param wantedFeilds
 194    *          the available table columns in no particular order
 195    * @return the pdb id field column index
 196    */
 197   public static int getPDBIdColumIndex(
 198           Collection<PDBDocField> wantedFeilds, boolean hasRefSeq)
 199   {
 200     int pdbFeildIndexCounter = hasRefSeq ? 1 : 0; // If a reference sequence is
 201                                            // attached then start counting from
 202                                            // 1 else start from zero
 203     for (PDBDocField feild : wantedFeilds)
 204     {
 205       if (feild.equals(PDBDocField.PDB_ID))
 206       {
 207         break; // once PDB Id index is determined exit iteration
 208       }
 209       ++pdbFeildIndexCounter;
 210     }
 211     return pdbFeildIndexCounter;
 212   }
 213
 214   /**
 215    * Represents the fields retrievable from a PDB Document response
 216    *
 217    */
 218   public enum PDBDocField
 219   {
 220     PDB_ID("PDB Id", "pdb_id"), TITLE("Title", "title"), MOLECULE_NAME(
 221             "Molecule", "molecule_name"), MOLECULE_TYPE("Molecule Type",
 222             "molecule_type"), MOLECULE_SEQUENCE("Sequence",
 223             "molecule_sequence"), PFAM_ACCESSION("PFAM Accession",
 224             "pfam_accession"), PFAM_NAME("PFAM Name", "pfam_name"), INTERPRO_NAME(
 225             "InterPro Name", "interpro_name"), INTERPRO_ACCESSION(
 226             "InterPro Accession", "interpro_accession"), UNIPROT_ID(
 227             "UniProt Id", "uniprot_id"), UNIPROT_ACCESSION(
 228             "UniProt Accession", "uniprot_accession"), UNIPROT_COVERAGE(
 229             "UniProt Coverage", "uniprot_coverage"), UNIPROT_FEATURES(
 230             "Uniprot Features", "uniprot_features"), R_FACTOR("R Factor",
 231             "r_factor"), RESOLUTION("Resolution", "resolution"), DATA_QUALITY(
 232             "Data Quality", "data_quality"), OVERALL_QUALITY(
 233             "Overall Quality", "overall_quality"), POLYMER_COUNT(
 234             "Number of Polymers", "number_of_polymers"), PROTEIN_CHAIN_COUNT(
 235             "Number of Protein Chains", "number_of_protein_chains"), BOUND_MOLECULE_COUNT(
 236             "Number of Bound Molecule", "number_of_bound_molecules"), POLYMER_RESIDUE_COUNT(
 237             "Number of Polymer Residue", "number_of_polymer_residues"), GENUS(
 238             "GENUS", "genus"), GENE_NAME("Gene Name", "gene_name"), EXPERIMENTAL_METHOD(
 239             "Experimental Method", "experimental_method"), GO_ID("GO Id",
 240             "go_id"), ASSEMBLY_ID("Assembly Id", "assembly_form"), ASSEMBLY_FORM(
 241             "Assembly Form", "assembly_id"), ASSEMBLY_TYPE("Assembly Type",
 242             "assembly_type"), SPACE_GROUP("Space Group", "spacegroup"), CATH_CODE(
 243             "Cath Code", "cath_code"), TAX_ID("Tax Id", "tax_id"), TAX_QUERY(
 244             "Tax Query", "tax_query"), INTERACTING_ENTRY_ID(
 245             "Interacting Entry Id", "interacting_entry_id"), INTERACTING_ENTITY_ID(
 246             "Interacting Entity Id", "interacting_entity_id"), INTERACTING_MOLECULES(
 247             "Interacting Molecules", "interacting_molecules"), PUBMED_ID(
 248             "Pubmed Id", "pubmed_id"), STATUS("Status", "status"), MODEL_QUALITY(
 249             "Model Quality", "model_quality"), PIVOT_RESOLUTION(
 250             "Pivot Resolution", "pivot_resolution"), DATA_REDUCTION_SOFTWARE(
 251             "Data reduction software", "data_reduction_software"), MAX_OBSERVED_RES(
 252             "Max observed residues", "max_observed_residues"), ORG_SCI_NAME(
 253             "Organism scientific name", "organism_scientific_name"), SUPER_KINGDOM(
 254             "Super kingdom", "superkingdom"), RANK("Rank", "rank"), CRYSTALLISATION_PH(
 255             "Crystallisation Ph", "crystallisation_ph"), BIO_FUNCTION(
 256             "Biological Function", "biological_function"), BIO_PROCESS(
 257             "Biological Process", "biological_process"), BIO_CELL_COMP(
 258             "Biological Cell Component", "biological_cell_component"), COMPOUND_NAME(
 259             "Compound Name", "compound_name"), COMPOUND_ID("Compound Id",
 260             "compound_id"), COMPOUND_WEIGHT("Compound Weight",
 261             "compound_weight"), COMP_SYS_NAME("Compound Systematic Name",
 262             "compound_systematic_name"), INTERACTING_LIG(
 263             "Interacting Ligands", "interacting_ligands"), JOURNAL(
 264             "Journal", "journal"), ALL_AUTHORS("All Authors", "all_authors"), EXPERIMENTAL_DATA_AVAILABLE(
 265             "Experiment Data Available", "experiment_data_available"), DIFFRACTION_PROTOCOL(
 266             "Diffraction Protocol", "diffraction_protocol"), REFINEMENT_SOFTWARE(
 267             "Refinement Software", "refinement_software"), STRUCTURE_DETERMINATION_METHOD(
 268             "Structure Determination Method",
 269             "structure_determination_method"), SYNCHROTON_SITE(
 270             "Synchrotron Site", "synchrotron_site"), SAMPLE_PREP_METHOD(
 271             "Sample Preparation Method", "sample_preparation_method"), ENTRY_AUTHORS(
 272             "Entry Authors", "entry_authors"), CITATION_TITLE(
 273             "Citation Title", "citation_title"), STRUCTURE_SOLUTION_SOFTWARE(
 274             "Structure Solution Software", "structure_solution_software"), ENTRY_ENTITY(
 275             "Entry Entity", "entry_entity"), R_FREE("R Free", "r_free"), NO_OF_POLYMER_ENTITIES(
 276             "Number of Polymer Entities", "number_of_polymer_entities"), NO_OF_BOUND_ENTITIES(
 277             "Number of Bound Entities", "number_of_bound_entities"), CRYSTALLISATION_RESERVOIR(
 278             "Crystallisation Reservoir", "crystallisation_reservoir"), DATA_SCALING_SW(
 279             "Data Scalling Software", "data_scaling_software"), DETECTOR(
 280             "Detector", "detector"), DETECTOR_TYPE("Detector Type",
 281             "detector_type"), MODIFIED_RESIDUE_FLAG(
 282             "Modified Residue Flag", "modified_residue_flag"), NUMBER_OF_COPIES(
 283             "Number of Copies", "number_of_copies"), STRUCT_ASYM_ID(
 284             "Struc Asym Id", "struct_asym_id"), HOMOLOGUS_PDB_ENTITY_ID(
 285             "Homologus PDB Entity Id", "homologus_pdb_entity_id"), MOLECULE_SYNONYM(
 286             "Molecule Synonym", "molecule_synonym"), DEPOSITION_SITE(
 287             "Deposition Site", "deposition_site"), SYNCHROTRON_BEAMLINE(
 288             "Synchrotron Beamline", "synchrotron_beamline"), ENTITY_ID(
 289             "Entity Id", "entity_id"), BEAM_SOURCE_NAME("Beam Source Name",
 290             "beam_source_name"), PROCESSING_SITE("Processing Site",
 291             "processing_site"), ENTITY_WEIGHT("Entity Weight",
 292             "entity_weight"), VERSION("Version", "_version_"), ALL("ALL",
 293             "text");
 294
 295     private String name;
 296
 297     private String code;
 298
 299     PDBDocField(String name, String code)
 300     {
 301       this.name = name;
 302       this.code = code;
 303     }
 304
 305     public String getName()
 306     {
 307       return name;
 308     }
 309
 310     public String getCode()
 311     {
 312       return code;
 313     }
 314
 315     public String toString()
 316     {
 317       return name;
 318     }
 319   }
 320 }