NCBI NETWORK SERVICES FOR ENTREZ QUERIES AND SEQUENCE FETCHING NCBI has new network services for Entrez (term lists, Boolean queries, document summaries, neighbors and links) and PubMed and Sequence record fetching. These are all CGI-based, were written by the group that maintains our web services, and have a lot of advantages over our older network services. The code is all in the ncbi/access folder, and is compiled into the ncbiobj library, along with the object loaders and api functions. This library is always linked in with our applications, and does not depend upon netcli or netentr libraries. The network connection functions are in the ncbi and ncbiconn libraries, also always linked in. The header files are ent2api.h for Entrez2 services and pmfapi.h for PubMed and sequence fetching. These functions are now accessed as named services, so the underlying URL is not present in the client code. This allows us to redirect requests between multiple servers for load balancing, and to change minor URL parameters as needed without breaking the client. The only parameters the client sends are essential ones (e.g., gi number). These services will thus be stable and supported over time, regardless of what we do behind the scenes. Another advantage is that the services can be called either synchronously or asynchronously. Synchronous queries wait for the reply before returning control back to your program, and are easier to code. For example: PubmedEntryPtr pep = PubMedSynchronousQuery (pmid); SeqEntryPtr sep = PubSeqSynchronousQuery (gi, 0, 0); These do not automatically try to reconnect if the network request failed, and the client can decide whether to put these calls in a small retry loop. Asynchronous queries are slighly more complicated to code, but allow better responsiveness, without the need for multi-threading in interactive programs. Asynchronous calls send the request to the server, add a block of data with the connection and your callback function to a queue, and immediately return control to your program. You are responsible for calling a queue checking function every so often, typically on a timer. When the results are ready, the queue checker calls the callback you specified. Your callback reads the data, does whatever you want with it, and returns to the queue checker, which cleans up the connection before returning to the timer event loop. Some NCBI toolkit library functions require the ability to fetch sequences from a registered service. For this, you can call PubSeqFetchEnable (); to enable access and PubSeqFetchDisable (); to disable access. The asn2gb -r flag (for remote access) works this way. The demo/entrez2 application uses desktop/e2trmlst.c and desktop/e2docsum.c to access the new servers. Some examples of how to use the new functions are provided below. /* standard startup */ #include #include #include #include static void StandardStartup (void) { ErrSetFatalLevel (SEV_MAX); ErrClearOptFlags (EO_SHOW_USERSTR); UseLocalAsnloadDataAndErrMsg (); ErrPathReset (); AllObjLoad (); SubmitAsnLoad (); FeatDefSetLoad (); SeqCodeSetLoad (); GeneticCodeTableLoad (); } /* enabling automatic fetch of components from SeqFetch service */ #include static void OnStartup (void) { PubSeqFetchEnable (); } static void OnShutdown (void) { PubSeqFetchDisable (); } /* example of fetching PubMed record */ #include #include static void SavePubMedRecord (Int4 pmid, FILE *fp) { PubmedEntryPtr pep; pep = PubMedSynchronousQuery (pmid); if (pep == NULL) return; MedlineEntryToDataFile ((MedlineEntryPtr) pep->medent, fp); PubmedEntryFree (pep); } /* example of fetching sequence record */ #include #include static void SaveSeqRecord (Int4 gi, FILE *fp) { SeqEntryPtr sep; sep = PubSeqSynchronousQuery (gi, 0, 0); if (sep == NULL) return; SeqEntryToGnbk (sep, NULL, GENBANK_FMT, RELEASE_MODE, NORMAL_STYLE, 0, 0, 0, NULL, fp); SeqEntryFree (sep); } /* example of Entrez2 Boolean query */ #include static void DoBooleanRequest (FILE *fp) { Entrez2BooleanReplyPtr e2br; Entrez2IdListPtr e2id; Entrez2RequestPtr e2rq; Entrez2ReplyPtr e2ry; Int4 i; Int4Ptr uids; e2rq = EntrezCreateBooleanRequest (TRUE, FALSE, "Nucleotide", NULL, 0, 0, NULL, 0, 0); if (e2rq == NULL) return; EntrezAddToBooleanRequest (e2rq, NULL, 0, "ORGN", "Saccharomyces cerevisiae", NULL, 0, 0, NULL, NULL, TRUE, TRUE); EntrezAddToBooleanRequest (e2rq, NULL, ENTREZ_OP_AND, NULL, NULL, NULL, 0, 0, NULL, NULL, TRUE, TRUE); EntrezAddToBooleanRequest (e2rq, NULL, 0, "PROP", "biomol mrna", NULL, 0, 0, NULL, NULL, TRUE, TRUE); e2ry = EntrezSynchronousQuery (e2rq); e2rq = Entrez2RequestFree (e2rq); if (e2ry == NULL) return; e2br = EntrezExtractBooleanReply (e2ry); if (e2br == NULL) return; if (e2br->count > 0) { e2id = e2br->uids; if (e2id != NULL && e2id->num > 0 && e2id->uids != NULL) { fprintf (fp, "count=%ld\n", (long) e2id->num); uids = (Int4Ptr) BSMerge (e2id->uids, NULL); if (uids != NULL) { for (i = 0; i < e2id->num; i++) { fprintf (fp, "pmid=%ld\n", (long) uids [i]); } MemFree (uids); } } } Entrez2BooleanReplyFree (e2br); } /* example of fetching Entrez2 document summary */ #include static void DoDocsumRequest (Int4 pmid, FILE *fp) { Entrez2DocsumDataPtr e2ddp; Entrez2DocsumListPtr e2dlp; Entrez2DocsumPtr e2dp; Entrez2RequestPtr e2rq; Entrez2ReplyPtr e2ry; e2rq = EntrezCreateDocSumRequest ("PubMed", pmid, 0, NULL, NULL); if (e2rq == NULL) return; e2ry = EntrezSynchronousQuery (e2rq); e2rq = Entrez2RequestFree (e2rq); if (e2ry == NULL) return; e2dlp = EntrezExtractDocsumReply (e2ry); if (e2dlp == NULL) return; for (e2dp = e2dlp->list; e2dp != NULL; e2dp = e2dp->next) { for (e2ddp = e2dp->docsum_data; e2ddp != NULL; e2ddp = e2ddp->next) { if (StringHasNoText (e2ddp->field_name)) continue; if (StringHasNoText (e2ddp->field_value)) continue; fprintf (fp, "%s - %s\n", e2ddp->field_name, e2ddp->field_value); } } Entrez2DocsumListFree (e2dlp); }