4418dae2476e397d9c03c1301e573e2c45b1a58e hiram Fri Feb 10 12:16:43 2023 -0800 adding asmEquivalent functions to this library no redmine diff --git src/hg/lib/asmEquivalent.c src/hg/lib/asmEquivalent.c index 64896b4..3971aadd 100644 --- src/hg/lib/asmEquivalent.c +++ src/hg/lib/asmEquivalent.c @@ -1,25 +1,27 @@ /* asmEquivalent.c was originally generated by the autoSql program, which also * generated asmEquivalent.h and asmEquivalent.sql. This module links the database and * the RAM representation of objects. */ #include "common.h" #include "linefile.h" #include "dystring.h" #include "jksql.h" #include "asmEquivalent.h" +#include "hdb.h" + char *asmEquivalentCommaSepFieldNames = "source,destination,sourceAuthority,destinationAuthority,matchCount,sourceCount,destinationCount"; /* definitions for sourceAuthority column */ static char *values_sourceAuthority[] = {"ensembl", "ucsc", "genbank", "refseq", NULL}; static struct hash *valhash_sourceAuthority = NULL; /* definitions for destinationAuthority column */ static char *values_destinationAuthority[] = {"ensembl", "ucsc", "genbank", "refseq", NULL}; static struct hash *valhash_destinationAuthority = NULL; void asmEquivalentStaticLoad(char **row, struct asmEquivalent *ret) /* Load a row from asmEquivalent table into ret. The contents of ret will * be replaced at the next call to this function. */ @@ -52,31 +54,31 @@ slAddHead(&list, el); } slReverse(&list); sqlFreeResult(&sr); return list; } void asmEquivalentSaveToDb(struct sqlConnection *conn, struct asmEquivalent *el, char *tableName, int updateSize) /* Save asmEquivalent as a row to the table specified by tableName. * As blob fields may be arbitrary size updateSize specifies the approx size * of a string that would contain the entire query. Arrays of native types are * converted to comma separated strings and loaded as such, User defined types are * inserted as NULL. This function automatically escapes quoted strings for mysql. */ { struct dyString *update = dyStringNew(updateSize); -sqlDyStringPrintf(update, "insert into %s values ( '%s','%s',(null),(null),%lld,%lld,%lld)", +sqlDyStringPrintf(update, "insert into %s values ( '%s','%s',%d,%d,%lld,%lld,%lld)", tableName, el->source, el->destination, el->sourceAuthority, el->destinationAuthority, el->matchCount, el->sourceCount, el->destinationCount); sqlUpdate(conn, update->string); dyStringFree(&update); } struct asmEquivalent *asmEquivalentLoad(char **row) /* Load a asmEquivalent from row fetched with select * from asmEquivalent * from database. Dispose of this with asmEquivalentFree(). */ { struct asmEquivalent *ret; AllocVar(ret); ret->source = cloneString(row[0]); ret->destination = cloneString(row[1]); ret->sourceAuthority = sqlEnumParse(row[2], values_sourceAuthority, &valhash_sourceAuthority); @@ -241,15 +243,65 @@ fprintf(f,"sourceCount"); fputc('"',f); fputc(':',f); fprintf(f, "%lld", el->sourceCount); fputc(',',f); fputc('"',f); fprintf(f,"destinationCount"); fputc('"',f); fputc(':',f); fprintf(f, "%lld", el->destinationCount); fputc('}',f); } /* -------------------------------- End autoSql Generated Code -------------------------------- */ +char *asmEquivalentUcscToNCBI(char *ucscName) +/* check if there is a RefSeq/GenBank equivalent to this UCSC assembly name. + * If RefSeq exists, return that first, else if GenBank than return that. + * No checking of sequence match counts in this first implementation, + * therefore, could be a fuzzy match, and since it is returning only the + * first one, it might not be the best match. Could add more specifics + * later to get better match. + */ +{ +char *ret = NULL; +if (ucscName == NULL) + return ret; + +struct sqlConnection *conn = hAllocConn("hgFixed"); +if (!conn) + return ret; + +if (!sqlTableExists(conn, "asmEquivalent")) + { + hFreeConn(&conn); + return ret; + } + +char buffer[4096]; + +sqlSafef(buffer, sizeof buffer, "SELECT destination FROM asmEquivalent WHERE sourceAuthority='ucsc' AND destinationAuthority='refseq' AND source='%s' LIMIT 1", ucscName); +char *sqlAnswer = sqlQuickString(conn, buffer); +if (isEmpty(sqlAnswer)) + { + sqlSafef(buffer, sizeof buffer, "SELECT destination FROM asmEquivalent WHERE sourceAuthority='ucsc' AND destinationAuthority='genbank' AND source='%s' LIMIT 1", ucscName); + sqlAnswer = sqlQuickString(conn, buffer); + } +hFreeConn(&conn); + +/* if there is a result, for example: GCA_000001405.28_GRCh38.p13 + * want to return only: GCA_000001405.28 + */ +if (isNotEmpty(sqlAnswer)) + { + char *words[3]; + int wordCount = 0; + wordCount = chopString(sqlAnswer, "_", words, ArraySize(words)); + if (3 != wordCount) /* something wrong with this answer */ + return ret; + safef(buffer, sizeof(buffer), "%s_%s", words[0], words[1]); + ret = cloneString(buffer); + } + +return ret; +}