38b9ede07fb3bfb98c74f611c62146b820747e4a braney Sat Dec 11 08:32:34 2021 -0800 check in custom hgc code for support of TOGA gene annotation from Michael Hiller's group diff --git src/hg/hgc/togaClick.c src/hg/hgc/togaClick.c new file mode 100644 index 0000000..86b49db --- /dev/null +++ src/hg/hgc/togaClick.c @@ -0,0 +1,371 @@ +/* togaClick - click handling for TOGA tracks */ +#include "common.h" +#include "hgc.h" +#include "togaClick.h" +#include "string.h" +#include "htmshell.h" + + +struct togaData *togaDataLoad(char **row) +/* Load a togaData from row fetched with select * from togaData + * from database. Dispose of this with togaDataFree(). */ +{ + struct togaData *ret; + AllocVar(ret); + ret->projection = cloneString(row[0]); + ret->ref_trans_id = cloneString(row[1]); + ret->ref_region = cloneString(row[2]); + ret->query_region = cloneString(row[3]); + ret->chain_score = cloneString(row[4]); + + ret->chain_synteny = cloneString(row[5]); + ret->chain_flank = cloneString(row[6]); + ret->chain_gl_cds_fract = cloneString(row[7]); + ret->chain_loc_cds_fract = cloneString(row[8]); + ret->chain_exon_cov = cloneString(row[9]); + + ret->chain_intron_cov = cloneString(row[10]); + ret->status = cloneString(row[11]); + ret->perc_intact_ign_M = cloneString(row[12]); + ret->perc_intact_int_M = cloneString(row[13]); + ret->intact_codon_prop = cloneString(row[14]); + + ret->ouf_prop = cloneString(row[15]); + ret->mid_intact = cloneString(row[16]); + ret->mid_pres = cloneString(row[17]); + ret->prot_alignment = cloneString(row[18]); + ret->svg_line = cloneString(row[19]); + return ret; +} + + +void togaDataFree(struct togaData **pEl) +/* Free a single dynamically allocated togaDatasuch as created + * with togaDataLoad(). */ +{ + struct togaData *el; + + if ((el = *pEl) == NULL) return; + freeMem(el->projection); + freeMem(el->ref_trans_id); + freeMem(el->ref_region); + freeMem(el->query_region); + freeMem(el->chain_score); + + freeMem(el->chain_synteny); + freeMem(el->chain_flank); + freeMem(el->chain_gl_cds_fract); + freeMem(el->chain_loc_cds_fract); + freeMem(el->chain_exon_cov); + + freeMem(el->chain_intron_cov); + freeMem(el->status); + freeMem(el->perc_intact_ign_M); + freeMem(el->perc_intact_int_M); + freeMem(el->intact_codon_prop); + + freeMem(el->ouf_prop); + freeMem(el->mid_intact); + freeMem(el->mid_pres); + freeMem(el->prot_alignment); + freeMem(el->svg_line); + freez(pEl); +} + +struct togaNucl *togaNuclLoad(char **row) +/* Load a togaNucl from row fetched with select * from togaNucl + * from database. Dispose of this with togaNuclFree(). */ +{ + struct togaNucl *ret; + AllocVar(ret); + ret->transcript = cloneString(row[0]); + ret->exon_num = cloneString(row[1]); + ret->exon_region = cloneString(row[2]); + ret->pid = cloneString(row[3]); + ret->blosum = cloneString(row[4]); + + ret->gaps = cloneString(row[5]); + ret->ali_class = cloneString(row[6]); + ret->exp_region = cloneString(row[7]); + ret->in_exp_region = cloneString(row[8]); + ret->alignment = cloneString(row[9]); + return ret; +} + + +void togaNuclFree(struct togaNucl **pEl) +/* Free a single dynamically allocated togaNucl such as created + * with togaNuclLoad(). */ +{ + struct togaNucl *el; + + if ((el = *pEl) == NULL) return; + freeMem(el->transcript); + freeMem(el->exon_num); + freeMem(el->exon_region); + freeMem(el->pid); + freeMem(el->blosum); + + freeMem(el->gaps); + freeMem(el->ali_class); + freeMem(el->exp_region); + freeMem(el->in_exp_region); + freeMem(el->alignment); + freez(pEl); +} + + +struct togaInactMut *togaInactMutLoad(char **row) +/* Load a togaInactMut from row fetched with select * from togaInactMut + * from database. Dispose of this with togaInactMutFree(). */ +{ + struct togaInactMut *ret; + AllocVar(ret); + ret->transcript = cloneString(row[0]); + ret->exon_num = cloneString(row[1]); + ret->position = cloneString(row[2]); + ret->mut_class = cloneString(row[3]); + ret->mutation = cloneString(row[4]); + + ret->is_inact = cloneString(row[5]); + ret->mut_id = cloneString(row[6]); + return ret; +} + + +void togaInactMutFree(struct togaInactMut **pEl) +/* Free a single dynamically allocated togaInactMut such as created + * with togaInactMutLoad(). */ +{ + struct togaInactMut *el; + if ((el = *pEl) == NULL) return; + freeMem(el->transcript); + freeMem(el->exon_num); + freeMem(el->position); + freeMem(el->mut_class); + freeMem(el->mutation); + + freeMem(el->is_inact); + freeMem(el->mut_id); + freez(pEl); +} + + +void extractHLTOGAsuffix(char *suffix) +/* Extract suffix from TOGA table name. +Prefix must be HLTOGAannot */ +{ + int suff_len = strlen(suffix); + if (suff_len <= HLTOGA_BED_PREFIX_LEN) + // we cannot chop first PREFIX_LEN characters + { + // TODO: NOT SURE IF IT WORKS; but this must not happen + char empty[5] = { '\0' }; + strcpy(suffix, empty); + } else { + // just start the string 11 characters upstream + memmove(suffix, suffix + HLTOGA_BED_PREFIX_LEN, suff_len - HLTOGA_BED_PREFIX_LEN + 1); + } +} + + +void doHillerLabTOGAGene(struct trackDb *tdb, char *item, char *table_name) +/* Put up TOGA Gene track info. */ +{ + //int start = cartInt(cart, "o"); + char headerTitle[512]; + char suffix[512]; + strcpy(suffix, table_name); + extractHLTOGAsuffix(suffix); + safef(headerTitle, sizeof(headerTitle), "%s", item); + genericHeader(tdb, headerTitle); + printf("

TOGA gene annotation

\n"); + // htmlHorizontalLine(); + struct sqlConnection *conn = hAllocConn(database); + + // define TOGA table names: initate with pre-defined prefixes + char togaDataTableName[256]; + char togaNuclTableName[256]; + char togaInactMutTableName[256]; + strcpy(togaDataTableName, HLTOGA_DATA_PREFIX); + strcpy(togaNuclTableName, HLTOGA_NUCL_PREFIX); + strcpy(togaInactMutTableName, HLTOGA_INACT_PREFIX); + // add suffix + strcat(togaDataTableName, suffix); + strcat(togaNuclTableName, suffix); + strcat(togaInactMutTableName, suffix); + + + if (hTableExists(database, togaDataTableName)) + { + printf("

Projection %s


\n", item); + char query[256]; + struct sqlResult *sr = NULL; + char **row; + struct togaData *info = NULL; + + sqlSafef(query, sizeof(query), "select * from %s where transcript='%s'", togaDataTableName, item); + sr = sqlGetResult(conn, query); + + if ((row = sqlNextRow(sr)) != NULL) { + info = togaDataLoad(row); // parse sql output + // fill HTML template: + printf("Projected via: %s
", + info->ref_trans_id, info->ref_trans_id); + printf("Region in reference: %s
\n", info->ref_region); + printf("Region in query: %s
\n", info->query_region); + + printf("Projection class: %s
\n", info->status); + printf("Chain score: %s
\n", info->chain_score); + // list of chain features (for orthology classification) + printf("Show chain features for classification\n"); + printf("
\n"); + printf("\n
\n
\n"); + htmlHorizontalLine(); + + // show inact mut plot + printf("

Inactivating mutations plot

\n"); + printf("%s
\n", info->svg_line); + + // GLP features + printf("Show GLP features\n"); + printf("
\n"); + printf("\n
\n
\n"); + + // and show protein sequence + htmlHorizontalLine(); + printf("

Protein sequence


\n"); + printf("Show protein alignment\n"); + printf("
\n"); + printf("%s
\n", info->prot_alignment); + printf("
\n

\n"); + + // do not forget to free toga data struct + togaDataFree(&info); + } else { + // no data found, need to report this + printf("Not found data for %s\n", item); + } + sqlFreeResult(&sr); + } + + // show inactivating mutations if required + printf("

Inactivating mutations


\n"); + + if (hTableExists(database, togaInactMutTableName)) + { + char query[256]; + struct sqlResult *sr = NULL; + char **row; + sqlSafef(query, sizeof(query), "select * from %s where transcript='%s'", togaInactMutTableName, item); + sr = sqlGetResult(conn, query); + printf("Show inactivating mutations\n"); + printf("
\n"); + printf("\n"); // init table + printf("\n"); + printf("\n"); + while ((row = sqlNextRow(sr)) != NULL) + { + struct togaInactMut *info = NULL; + info = togaInactMutLoad(row); + printf("\n"); + printf("\n", info->exon_num); + printf("\n", info->position); + printf("\n", info->mut_class); + printf("\n", info->mutation); + if (sameWord(info->is_inact, ONE_)){ + printf("\n", YES_); + } else { + printf("\n", NO_); + } + printf("\n", info->mut_id); + printf("\n"); + togaInactMutFree(&info); + } + sqlFreeResult(&sr); + printf("
exonposm_classmutis_inactmut_id
%s%s%s%s%s%s%s
\n"); + printf("
\n
\n"); + } else { + printf("Sorry, cannot find TOGAInactMut table.
\n"); + } + + // show exons data + htmlHorizontalLine(); + printf("

Exons data


\n"); + + if (hTableExists(database, togaNuclTableName)) + { + char query[256]; + struct sqlResult *sr = NULL; + char **row; + printf("Show exon sequences and features\n"); + printf("
\n"); + sqlSafef(query, sizeof(query), "select * from %s where transcript='%s'", togaNuclTableName, item); + sr = sqlGetResult(conn, query); + + while ((row = sqlNextRow(sr)) != NULL) + { + struct togaNucl *info = NULL; + info = togaNuclLoad(row); + printf("
Exon number: %s

\n", info->exon_num); + printf("Exon region: %s
\n", info->exon_region); + printf("Nucleotide percent identity: %s | BLOSUM: %s
\n", info->pid, info->blosum); + if (sameWord(info->gaps, ONE_)){ + printf("Intersects assembly gaps: %s
\n", YES_); + } else { + printf("Intersects assembly gaps: %s
\n", NO_); + } + printf("Exon alignment class: %s
\n", info->ali_class); + if (sameWord(info->in_exp_region, ONE_)){ + printf("Detected within expected region: %s
\n", YES_); + } else { + printf("Detected within expected region: %s
\n", NO_); + } + printf("Expected region: %s
\n", info->exp_region); + printf("
\n"); + printf("Sequence alignment:
\n"); + printf("%s
\n", info->alignment); + togaNuclFree(&info); + } + sqlFreeResult(&sr); + printf("
\n

\n"); + } else { + printf("Sorry, cannot find TOGANucl table.
\n"); + } + + htmlHorizontalLine(); + + // TODO: check whether I need this + printf("%s", hgTracksPathAndSettings()); + hPrintf(""); + hPrintf(""); + hPrintf(""); + + + printTrackHtml(tdb); // and do I need this? + hFreeConn(&conn); +}