rc_t load_consensus_src( con_ctx * sctx, KDirectory * hdf5_src )
{
BaseCalls_cmn ConsensusTab;
rc_t rc = 0;
if ( sctx->lctx->check_src_obj )
rc = check_src_objects( hdf5_src, consensus_groups_to_check,
consensus_tables_to_check, false );
if ( rc == 0 )
rc = open_BaseCalls_cmn( hdf5_src, &ConsensusTab, true,
"PulseData/ConsensusBaseCalls", sctx->lctx->cache_content, true );
if ( rc == 0 )
{
uint64_t total_bases = zmw_total( &ConsensusTab.zmw );
uint64_t total_spots = ConsensusTab.zmw.NumEvent.extents[ 0 ];
KLogLevel tmp_lvl = KLogLevelGet();
KLogLevelSet( klogInfo );
PLOGMSG( klogInfo, ( klogInfo,
"loading consensus-table ( $(bases) bases / $(spots) spots ):",
"bases=%lu,spots=%lu", total_bases, total_spots ));
KLogLevelSet( tmp_lvl );
if ( !check_Consensus_totalcount( &ConsensusTab, total_bases ) )
rc = RC( rcExe, rcNoTarg, rcAllocating, rcParam, rcInvalid );
else
rc = zmw_for_each( &ConsensusTab.zmw, &sctx->lctx->xml_progress, sctx->cursor,
sctx->lctx->with_progress, sctx->col_idx, NULL,
true, consensus_load_spot, &ConsensusTab );
close_BaseCalls_cmn( &ConsensusTab );
}
return rc;
}
static rc_t consensus_loader( ld_context *lctx, const char * table_name,
bool cache_content )
{
uint32_t col_idx[ consensus_tab_count ];
rc_t rc = add_columns( lctx->cursor, consensus_tab_count, -1,
col_idx, consensus_tab_names );
if ( rc == 0 )
{
rc = VCursorOpen( lctx->cursor );
if ( rc != 0 )
LOGERR( klogErr, rc, "cannot open cursor on consensus-table" );
else
{
BaseCalls_cmn ConsensusTab;
const INSDC_SRA_platform_id platform = SRA_PLATFORM_PACBIO_SMRT;
rc = VCursorDefault ( lctx->cursor, col_idx[ consensus_tab_PLATFORM ],
sizeof platform * 8, &platform, 0, 1 );
if ( rc != 0 )
LOGERR( klogErr, rc, "cannot set cursor-default on consensus-table for platform-column" );
else
{
const INSDC_SRA_read_filter filter = SRA_READ_FILTER_PASS;
rc = VCursorDefault ( lctx->cursor, col_idx[ consensus_tab_READ_FILTER ],
sizeof filter * 8, &filter, 0, 1 );
if ( rc != 0 )
LOGERR( klogErr, rc, "cannot set cursor-default on consensus-table for read-filter-column" );
}
if ( rc == 0 )
rc = open_BaseCalls_cmn( lctx->hdf5_dir, &ConsensusTab, true,
"PulseData/ConsensusBaseCalls", cache_content, true );
if ( rc == 0 )
{
uint64_t total_bases = zmw_total( &ConsensusTab.zmw );
uint64_t total_spots = ConsensusTab.zmw.NumEvent.extents[ 0 ];
KLogLevel tmp_lvl = KLogLevelGet();
KLogLevelSet( klogInfo );
PLOGMSG( klogInfo, ( klogInfo,
"loading consensus-table ( $(bases) bases / $(spots) spots ):",
"bases=%lu,spots=%lu", total_bases, total_spots ));
KLogLevelSet( tmp_lvl );
if ( check_Consensus_totalcount( &ConsensusTab, total_bases ) )
rc = zmw_for_each( &ConsensusTab.zmw, lctx, col_idx, NULL,
true, consensus_load_spot, &ConsensusTab );
else
rc = RC( rcExe, rcNoTarg, rcAllocating, rcParam, rcInvalid );
close_BaseCalls_cmn( &ConsensusTab );
}
}
}
return rc;
}
rc_t load_seq_src( seq_ctx * sctx, KDirectory * hdf5_src )
{
rc_t rc = 0;
if ( sctx->lctx->check_src_obj )
rc = check_src_objects( hdf5_src, seq_groups_to_check, seq_tables_to_check, true );
if ( rc == 0 && !sctx->src_open )
rc = open_BaseCalls( hdf5_src, &sctx->BaseCallsTab, "PulseData/BaseCalls", sctx->lctx->cache_content, &sctx->rgn_present );
if ( rc == 0 )
{
/* calculates the total number of bases, according to the zmw-table */
uint64_t total_bases = zmw_total( &sctx->BaseCallsTab.cmn.zmw );
/* calculates the total number of spots, according to the zmw-table */
uint64_t total_spots = sctx->BaseCallsTab.cmn.zmw.NumEvent.extents[ 0 ];
KLogLevel tmp_lvl = KLogLevelGet();
KLogLevelSet( klogInfo );
PLOGMSG( klogInfo, ( klogInfo,
"loading sequence-table ( $(bases) bases / $(spots) spots ):",
"bases=%lu,spots=%lu", total_bases, total_spots ));
KLogLevelSet( tmp_lvl );
/* checks that all tables, which are loaded do have the correct
number of values (the number that the zmw-table requests) */
if ( !check_BaseCall_totalcount( &sctx->BaseCallsTab, total_bases ) )
rc = RC( rcExe, rcNoTarg, rcAllocating, rcParam, rcInvalid );
else
{
region_type_mapping mapping;
if ( sctx->rgn_present )
{
const KNamelist *region_types;
/* read the meta-data-entry "RegionTypes" of the hdf5-regions-table
into a KNamelist */
rc = KArrayFileGetMeta ( sctx->BaseCallsTab.rgn.hdf5_regions.af, "RegionTypes", ®ion_types );
if ( rc != 0 )
{
LOGERR( klogErr, rc, "cannot read Regions.RegionTypes" );
}
else
{
/* extract the region-type-mapping out of the read KNamelist */
rc = rgn_extract_type_mappings( region_types, &mapping, false );
KNamelistRelease ( region_types );
if ( rc != 0 )
{
LOGERR( klogErr, rc, "cannot map regions-types" );
}
}
}
if ( rc == 0 )
{
region_type_mapping *mapping_ptr = NULL;
if ( sctx->rgn_present )
mapping_ptr = &mapping;
/* call for every spot the function >seq_load_spot< */
rc = zmw_for_each( &sctx->BaseCallsTab.cmn.zmw, &sctx->lctx->xml_progress, sctx->cursor,
sctx->lctx->with_progress, sctx->col_idx, mapping_ptr, false,
seq_load_spot, &sctx->BaseCallsTab );
}
if ( sctx->rgn_present )
seq_load_info( &sctx->BaseCallsTab.rgn.stat );
sctx->lctx->total_seq_bases += total_bases;
sctx->lctx->total_seq_spots += total_spots;
}
close_BaseCalls( &sctx->BaseCallsTab );
sctx->src_open = false;
}
return rc;
}
static rc_t seq_loader( ld_context *lctx, KDirectory * hdf5_src, VCursor * cursor, const char * table_name )
{
BaseCalls BaseCallsTab;
bool rgn_present;
/* opens all hdf5-tables, which are needed to load the sequence-table */
rc_t rc = open_BaseCalls( hdf5_src, &BaseCallsTab,
"PulseData/BaseCalls", lctx->cache_content, &rgn_present );
if ( rc == 0 )
{
/* calculates the total number of bases, according to the zmw-table */
uint64_t total_bases = zmw_total( &BaseCallsTab.cmn.zmw );
/* calculates the total number of spots, according to the zmw-table */
uint64_t total_spots = BaseCallsTab.cmn.zmw.NumEvent.extents[ 0 ];
KLogLevel tmp_lvl = KLogLevelGet();
KLogLevelSet( klogInfo );
PLOGMSG( klogInfo, ( klogInfo,
"loading sequence-table ( $(bases) bases / $(spots) spots ):",
"bases=%lu,spots=%lu", total_bases, total_spots ));
KLogLevelSet( tmp_lvl );
/* checks that all tables, which are loaded do have the correct
number of values (the number that the zmw-table requests) */
if ( check_BaseCall_totalcount( &BaseCallsTab, total_bases ) )
{
region_type_mapping mapping;
if ( rgn_present )
{
const KNamelist *region_types;
/* read the meta-data-entry "RegionTypes" of the hdf5-regions-table
into a KNamelist */
rc = KArrayFileGetMeta ( BaseCallsTab.rgn.hdf5_regions.af, "RegionTypes", ®ion_types );
if ( rc != 0 )
{
LOGERR( klogErr, rc, "cannot read Regions.RegionTypes" );
}
else
{
/* extract the region-type-mapping out of the read KNamelist */
rc = rgn_extract_type_mappings( region_types, &mapping, false );
KNamelistRelease ( region_types );
if ( rc != 0 )
{
LOGERR( klogErr, rc, "cannot map regions-types" );
}
}
}
if ( rc == 0 )
{
/* holds the vdb-column-index for every column we write */
uint32_t col_idx[ seq_tab_count ];
int32_t to_exclude;
/* depending on the bit-size of the PulseIndex-table in HDF5
exclude the opposite column from VDB */
if ( BaseCallsTab.PulseIndex.element_bits == PULSE_INDEX_BITSIZE_16 )
to_exclude = seq_tab_PULSE_INDEX_32;
else
to_exclude = seq_tab_PULSE_INDEX_16;
/* add all columns to the vdb-cursor */
rc = add_columns( cursor, seq_tab_count, to_exclude, col_idx, seq_tab_names );
if ( rc == 0 )
{
rc = VCursorOpen( cursor );
if ( rc != 0 )
LOGERR( klogErr, rc, "cannot open cursor on seq-table" );
else
{
const uint8_t platform = SRA_PLATFORM_PACBIO_SMRT;
rc = VCursorDefault ( cursor, col_idx[ seq_tab_PLATFORM ], 8, &platform, 0, 1 );
if ( rc != 0 )
LOGERR( klogErr, rc, "cannot set cursor-default on seq-table for platform-column" );
else
{
region_type_mapping *mapping_ptr = NULL;
if ( rgn_present )
{
mapping_ptr = &mapping;
}
/* call for every spot the function >seq_load_spot< */
rc = zmw_for_each( &BaseCallsTab.cmn.zmw, &lctx->xml_progress, cursor,
lctx->with_progress, col_idx, mapping_ptr, false, seq_load_spot, &BaseCallsTab );
}
}
}
}
if ( rgn_present )
{
seq_load_info( &BaseCallsTab.rgn.stat );
if ( mapping.count_of_unknown_rgn_types > 0 )
{
KLogLevel tmp_lvl = KLogLevelGet();
KLogLevelSet( klogInfo );
PLOGMSG( klogInfo, ( klogInfo,
"$(times) x unknown region types encountered",
"times=%i", mapping.count_of_unknown_rgn_types ) );
KLogLevelSet( tmp_lvl );
}
}
lctx->total_seq_bases += total_bases;
lctx->total_seq_spots += total_spots;
}
else
rc = RC( rcExe, rcNoTarg, rcAllocating, rcParam, rcInvalid );
close_BaseCalls( &BaseCallsTab );
}
return rc;
}