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 /**
  27  * A rest client for querying the Search endpoing of the PDB REST API
  28  *
  29  * @author tcnofoegbu
  30  *
  31  */
  32 public class PDBRestClient
  33 {
  34   private static String PDB_SEARCH_ENDPOINT = "http://www.ebi.ac.uk/pdbe/search/pdb/select?";
  35
  36   private static int DEFAULT_RESPONSE_SIZE = 200;
  37
  38   /**
  39    * Takes a PDBRestRequest object and returns a response upon execution
  40    *
  41    * @param pdbRestRequest
  42    *          the PDBRestRequest instance to be processed
  43    * @return the pdbResponse object for the given request
  44    */
  45   public PDBRestResponse executeRequest(PDBRestRequest pdbRestRequest)
  46   {
  47     ClientConfig clientConfig = new DefaultClientConfig();
  48     clientConfig.getFeatures().put(JSONConfiguration.FEATURE_POJO_MAPPING,
  49             Boolean.TRUE);
  50     Client client = Client.create(clientConfig);
  51
  52     String wantedFields = getPDBDocFieldsAsCommaDelimitedString(pdbRestRequest
  53             .getWantedFields());
  54     int responseSize = (pdbRestRequest.getResponseSize() == 0) ? DEFAULT_RESPONSE_SIZE
  55             : pdbRestRequest.getResponseSize();
  56     String sortParam = (pdbRestRequest.getFieldToSortBy() == null || pdbRestRequest
  57             .getFieldToSortBy().trim().isEmpty()) ? "" : (pdbRestRequest
  58             .getFieldToSortBy() + (pdbRestRequest.isAscending() ? " asc"
  59             : " desc"));
  60
  61     // Build request parameters for the REST Request
  62     WebResource webResource = client.resource(PDB_SEARCH_ENDPOINT)
  63             .queryParam("wt", "json").queryParam("fl", wantedFields)
  64             .queryParam("rows", String.valueOf(responseSize))
  65             .queryParam("q", pdbRestRequest.getQuery())
  66             .queryParam("sort", sortParam);
  67
  68     // Execute the REST request
  69     ClientResponse clientResponse = webResource.accept(
  70             MediaType.APPLICATION_JSON).get(ClientResponse.class);
  71
  72     // Get the JSON string from the response object
  73     String responseString = clientResponse.getEntity(String.class);
  74
  75     // Check the response status and report exception if one occurs
  76     if (clientResponse.getStatus() != 200)
  77     {
  78       String errorMessage = "";
  79       if (clientResponse.getStatus() == 400)
  80       {
  81         errorMessage = parseJsonExceptionString(responseString);
  82         throw new RuntimeException(errorMessage);
  83       }
  84       else
  85       {
  86         errorMessage = "Failed : HTTP error code : "
  87                 + clientResponse.getStatus();
  88         throw new RuntimeException(errorMessage);
  89       }
  90     }
  91
  92     // Make redundant objects eligible for garbage collection to conserve
  93     // memory
  94     clientResponse = null;
  95     client = null;
  96
  97     // Process the response and return the result to the caller.
  98     return parsePDBJsonResponse(responseString, pdbRestRequest);
  99   }
 100
 101   /**
 102    * Process error response from PDB server if/when one occurs.
 103    *
 104    * @param jsonResponse
 105    *          the JSON string containing error message from the server
 106    * @return the processed error message from the JSON string
 107    */
 108   public static String parseJsonExceptionString(String jsonErrorResponse)
 109   {
 110     StringBuilder errorMessage = new StringBuilder(
 111             "\n============= PDB Rest Client RunTime error =============\n");
 112
 113     try
 114     {
 115       JSONParser jsonParser = new JSONParser();
 116       JSONObject jsonObj = (JSONObject) jsonParser.parse(jsonErrorResponse);
 117       JSONObject errorResponse = (JSONObject) jsonObj.get("error");
 118
 119       JSONObject responseHeader = (JSONObject) jsonObj
 120               .get("responseHeader");
 121       JSONObject paramsObj = (JSONObject) responseHeader.get("params");
 122       String status = responseHeader.get("status").toString();
 123       String message = errorResponse.get("msg").toString();
 124       String query = paramsObj.get("q").toString();
 125       String fl = paramsObj.get("fl").toString();
 126
 127       errorMessage.append("Status: ").append(status).append("\n");
 128       errorMessage.append("Message: ").append(message).append("\n");
 129       errorMessage.append("query: ").append(query).append("\n");
 130       errorMessage.append("fl: ").append(fl).append("\n");
 131
 132     } catch (ParseException e)
 133     {
 134       e.printStackTrace();
 135     }
 136     return errorMessage.toString();
 137   }
 138
 139   /**
 140    * Parses the JSON response string from PDB REST API. The response is dynamic
 141    * hence, only fields specifically requested for in the 'wantedFields'
 142    * parameter is fetched/processed
 143    *
 144    * @param pdbJsonResponseString
 145    *          the JSON string to be parsed
 146    * @param pdbRestRequest
 147    *          the request object which contains parameters used to process the
 148    *          JSON string
 149    * @return
 150    */
 151   @SuppressWarnings("unchecked")
 152   public static PDBRestResponse parsePDBJsonResponse(
 153           String pdbJsonResponseString, PDBRestRequest pdbRestRequest)
 154   {
 155     PDBRestResponse searchResult = new PDBRestResponse();
 156     List<PDBResponseSummary> result = null;
 157     try
 158     {
 159       JSONParser jsonParser = new JSONParser();
 160       JSONObject jsonObj = (JSONObject) jsonParser
 161               .parse(pdbJsonResponseString);
 162
 163       JSONObject pdbResponse = (JSONObject) jsonObj.get("response");
 164       String queryTime = ((JSONObject) jsonObj.get("responseHeader")).get(
 165               "QTime").toString();
 166       int numFound = Integer
 167               .valueOf(pdbResponse.get("numFound").toString());
 168       if (numFound > 0)
 169       {
 170         result = new ArrayList<PDBResponseSummary>();
 171         JSONArray docs = (JSONArray) pdbResponse.get("docs");
 172         for (Iterator<JSONObject> docIter = docs.iterator(); docIter
 173                 .hasNext();)
 174         {
 175           JSONObject doc = docIter.next();
 176           result.add(searchResult.new PDBResponseSummary(doc,
 177                   pdbRestRequest));
 178         }
 179         searchResult.setNumberOfItemsFound(numFound);
 180         searchResult.setResponseTime(queryTime);
 181         searchResult.setSearchSummary(result);
 182       }
 183     } catch (ParseException e)
 184     {
 185       e.printStackTrace();
 186     }
 187     return searchResult;
 188   }
 189
 190   /**
 191    * Takes a collection of PDBDocField and converts its 'code' Field values into
 192    * a comma delimited string.
 193    *
 194    * @param pdbDocfields
 195    *          the collection of PDBDocField to process
 196    * @return the comma delimited string from the pdbDocFields collection
 197    */
 198   public static String getPDBDocFieldsAsCommaDelimitedString(
 199           Collection<PDBDocField> pdbDocfields)
 200   {
 201     String result = "";
 202     if (pdbDocfields != null && !pdbDocfields.isEmpty())
 203     {
 204       StringBuilder returnedFields = new StringBuilder();
 205       for (PDBDocField field : pdbDocfields)
 206       {
 207         returnedFields.append(",").append(field.getCode());
 208       }
 209       returnedFields.deleteCharAt(0);
 210       result = returnedFields.toString();
 211     }
 212     return result;
 213   }
 214
 215   /**
 216    * Determines the column index for 'PDB Id' Fields in the dynamic summary
 217    * table. The PDB Id serves as a unique identifier for a given row in the
 218    * summary table
 219    *
 220    * @param wantedFeilds
 221    *          the available table columns in no particular order
 222    * @return the pdb id field column index
 223    */
 224   public static int getPDBIdColumIndex(
 225           Collection<PDBDocField> wantedFeilds, boolean hasRefSeq)
 226   {
 227
 228     // If a reference sequence is attached then start counting from 1 else
 229     // start from zero
 230     int pdbFeildIndexCounter = hasRefSeq ? 1 : 0;
 231
 232     for (PDBDocField feild : wantedFeilds)
 233     {
 234       if (feild.equals(PDBDocField.PDB_ID))
 235       {
 236         break; // Once PDB Id index is determined exit iteration
 237       }
 238       ++pdbFeildIndexCounter;
 239     }
 240     return pdbFeildIndexCounter;
 241   }
 242
 243   /**
 244    * This enum represents the fields available in the PDB JSON response
 245    *
 246    */
 247   public enum PDBDocField
 248   {
 249     PDB_ID("PDB Id", "pdb_id"), TITLE("Title", "title"), MOLECULE_NAME(
 250             "Molecule", "molecule_name"), MOLECULE_TYPE("Molecule Type",
 251             "molecule_type"), MOLECULE_SEQUENCE("Sequence",
 252             "molecule_sequence"), PFAM_ACCESSION("PFAM Accession",
 253             "pfam_accession"), PFAM_NAME("PFAM Name", "pfam_name"), INTERPRO_NAME(
 254             "InterPro Name", "interpro_name"), INTERPRO_ACCESSION(
 255             "InterPro Accession", "interpro_accession"), UNIPROT_ID(
 256             "UniProt Id", "uniprot_id"), UNIPROT_ACCESSION(
 257             "UniProt Accession", "uniprot_accession"), UNIPROT_COVERAGE(
 258             "UniProt Coverage", "uniprot_coverage"), UNIPROT_FEATURES(
 259             "Uniprot Features", "uniprot_features"), R_FACTOR("R Factor",
 260             "r_factor"), RESOLUTION("Resolution", "resolution"), DATA_QUALITY(
 261             "Data Quality", "data_quality"), OVERALL_QUALITY(
 262             "Overall Quality", "overall_quality"), POLYMER_COUNT(
 263             "Number of Polymers", "number_of_polymers"), PROTEIN_CHAIN_COUNT(
 264             "Number of Protein Chains", "number_of_protein_chains"), BOUND_MOLECULE_COUNT(
 265             "Number of Bound Molecule", "number_of_bound_molecules"), POLYMER_RESIDUE_COUNT(
 266             "Number of Polymer Residue", "number_of_polymer_residues"), GENUS(
 267             "GENUS", "genus"), GENE_NAME("Gene Name", "gene_name"), EXPERIMENTAL_METHOD(
 268             "Experimental Method", "experimental_method"), GO_ID("GO Id",
 269             "go_id"), ASSEMBLY_ID("Assembly Id", "assembly_form"), ASSEMBLY_FORM(
 270             "Assembly Form", "assembly_id"), ASSEMBLY_TYPE("Assembly Type",
 271             "assembly_type"), SPACE_GROUP("Space Group", "spacegroup"), CATH_CODE(
 272             "Cath Code", "cath_code"), TAX_ID("Tax Id", "tax_id"), TAX_QUERY(
 273             "Tax Query", "tax_query"), INTERACTING_ENTRY_ID(
 274             "Interacting Entry Id", "interacting_entry_id"), INTERACTING_ENTITY_ID(
 275             "Interacting Entity Id", "interacting_entity_id"), INTERACTING_MOLECULES(
 276             "Interacting Molecules", "interacting_molecules"), PUBMED_ID(
 277             "Pubmed Id", "pubmed_id"), STATUS("Status", "status"), MODEL_QUALITY(
 278             "Model Quality", "model_quality"), PIVOT_RESOLUTION(
 279             "Pivot Resolution", "pivot_resolution"), DATA_REDUCTION_SOFTWARE(
 280             "Data reduction software", "data_reduction_software"), MAX_OBSERVED_RES(
 281             "Max observed residues", "max_observed_residues"), ORG_SCI_NAME(
 282             "Organism scientific name", "organism_scientific_name"), SUPER_KINGDOM(
 283             "Super kingdom", "superkingdom"), RANK("Rank", "rank"), CRYSTALLISATION_PH(
 284             "Crystallisation Ph", "crystallisation_ph"), BIOLOGICAL_FUNCTION(
 285             "Biological Function", "biological_function"), BIOLOGICAL_PROCESS(
 286             "Biological Process", "biological_process"), BIOLOGICAL_CELL_COMPONENT(
 287             "Biological Cell Component", "biological_cell_component"), COMPOUND_NAME(
 288             "Compound Name", "compound_name"), COMPOUND_ID("Compound Id",
 289             "compound_id"), COMPOUND_WEIGHT("Compound Weight",
 290             "compound_weight"), COMPOUND_SYSTEMATIC_NAME(
 291             "Compound Systematic Name", "compound_systematic_name"), INTERACTING_LIG(
 292             "Interacting Ligands", "interacting_ligands"), JOURNAL(
 293             "Journal", "journal"), ALL_AUTHORS("All Authors", "all_authors"), EXPERIMENTAL_DATA_AVAILABLE(
 294             "Experiment Data Available", "experiment_data_available"), DIFFRACTION_PROTOCOL(
 295             "Diffraction Protocol", "diffraction_protocol"), REFINEMENT_SOFTWARE(
 296             "Refinement Software", "refinement_software"), STRUCTURE_DETERMINATION_METHOD(
 297             "Structure Determination Method",
 298             "structure_determination_method"), SYNCHROTON_SITE(
 299             "Synchrotron Site", "synchrotron_site"), SAMPLE_PREP_METHOD(
 300             "Sample Preparation Method", "sample_preparation_method"), ENTRY_AUTHORS(
 301             "Entry Authors", "entry_authors"), CITATION_TITLE(
 302             "Citation Title", "citation_title"), STRUCTURE_SOLUTION_SOFTWARE(
 303             "Structure Solution Software", "structure_solution_software"), ENTRY_ENTITY(
 304             "Entry Entity", "entry_entity"), R_FREE("R Free", "r_free"), NO_OF_POLYMER_ENTITIES(
 305             "Number of Polymer Entities", "number_of_polymer_entities"), NO_OF_BOUND_ENTITIES(
 306             "Number of Bound Entities", "number_of_bound_entities"), CRYSTALLISATION_RESERVOIR(
 307             "Crystallisation Reservoir", "crystallisation_reservoir"), DATA_SCALING_SW(
 308             "Data Scalling Software", "data_scaling_software"), DETECTOR(
 309             "Detector", "detector"), DETECTOR_TYPE("Detector Type",
 310             "detector_type"), MODIFIED_RESIDUE_FLAG(
 311             "Modified Residue Flag", "modified_residue_flag"), NUMBER_OF_COPIES(
 312             "Number of Copies", "number_of_copies"), STRUCT_ASYM_ID(
 313             "Struc Asym Id", "struct_asym_id"), HOMOLOGUS_PDB_ENTITY_ID(
 314             "Homologus PDB Entity Id", "homologus_pdb_entity_id"), MOLECULE_SYNONYM(
 315             "Molecule Synonym", "molecule_synonym"), DEPOSITION_SITE(
 316             "Deposition Site", "deposition_site"), SYNCHROTRON_BEAMLINE(
 317             "Synchrotron Beamline", "synchrotron_beamline"), ENTITY_ID(
 318             "Entity Id", "entity_id"), BEAM_SOURCE_NAME("Beam Source Name",
 319             "beam_source_name"), PROCESSING_SITE("Processing Site",
 320             "processing_site"), ENTITY_WEIGHT("Entity Weight",
 321             "entity_weight"), VERSION("Version", "_version_"), ALL("ALL",
 322             "text");
 323
 324     private String name;
 325
 326     private String code;
 327
 328     PDBDocField(String name, String code)
 329     {
 330       this.name = name;
 331       this.code = code;
 332     }
 333
 334     public String getName()
 335     {
 336       return name;
 337     }
 338
 339     public String getCode()
 340     {
 341       return code;
 342     }
 343
 344     public String toString()
 345     {
 346       return name;
 347     }
 348   }
 349 }