obitools/obitools3
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 66 Wiki Activity

Alignment: updated functions to align columns (LCS)

Browse Source
This commit is contained in:
Celine Mercier
2018-11-07 16:00:58 +01:00
parent 6b1c41f3fb
commit d076ea9900
2 changed files with 160 additions and 74 deletions
Download Patch File Download Diff File Expand all files Collapse all files

208
src/obi_align.c → src/obi_lcs.c
View File

@ -3,7 +3,7 @@
****************************************************************************/ ****************************************************************************/
/** /**
* @file obi_align.c * @file obi_lcs.c
* @author Celine Mercier (celine.mercier@metabarcoding.org) * @author Celine Mercier (celine.mercier@metabarcoding.org)
* @date May 4th 2016 * @date May 4th 2016
* @brief Functions handling LCS sequence alignments. * @brief Functions handling LCS sequence alignments.
@ -18,7 +18,7 @@
#include <stdio.h> #include <stdio.h>
#include <stdbool.h> #include <stdbool.h>
#include "obi_align.h" #include "obi_lcs.h"
#include "obidebug.h" #include "obidebug.h"
#include "obierrno.h" #include "obierrno.h"
#include "obitypes.h" #include "obitypes.h"
@ -121,11 +121,11 @@ static int print_alignment_result(Obiview_p output_view,
OBIDMS_column_p seq2_column, OBIDMS_column_p seq2_column,
index_t seq1_idx, index_t seq1_idx,
index_t seq2_idx, index_t seq2_idx,
// bool print_count, bool print_count,
// OBIDMS_column_p count1_column, OBIDMS_column_p count1_column,
// OBIDMS_column_p count2_column, OBIDMS_column_p count2_column,
// int count1, int count1,
// int count2, int count2,
OBIDMS_column_p ali_length_column, OBIDMS_column_p ali_length_column,
int ali_length, int ali_length,
OBIDMS_column_p lcs_length_column, OBIDMS_column_p lcs_length_column,
@ -231,22 +231,22 @@ static int create_alignment_output_columns(Obiview_p output_view,
return -1; return -1;
} }
} }
// if (print_count) // TODO count columns not implemented yet if (print_count)
// { {
// // Create the column for the count of the first sequences aligned // Create the column for the count of the first sequences aligned
// if (obi_view_add_column(output_view, COUNT1_COLUMN_NAME, -1, NULL, OBI_INT, 0, 1, NULL, false, false, false, NULL, NULL, -1, COUNT1_COLUMN_COMMENTS, true) < 0) if (obi_view_add_column(output_view, COUNT1_COLUMN_NAME, -1, NULL, OBI_INT, 0, 1, NULL, false, false, false, NULL, NULL, -1, COUNT1_COLUMN_COMMENTS, true) < 0)
// { {
// obidebug(1, "\nError creating the first column for the sequence counts when aligning"); obidebug(1, "\nError creating the first column for the sequence counts when aligning");
// return -1; return -1;
// } }
//
// // Create the column for the count of the second sequences aligned // Create the column for the count of the second sequences aligned
// if (obi_view_add_column(output_view, COUNT2_COLUMN_NAME, -1, NULL, OBI_INT, 0, 1, NULL, false, false, false, NULL, NULL, -1, COUNT2_COLUMN_COMMENTS, true) < 0) if (obi_view_add_column(output_view, COUNT2_COLUMN_NAME, -1, NULL, OBI_INT, 0, 1, NULL, false, false, false, NULL, NULL, -1, COUNT2_COLUMN_COMMENTS, true) < 0)
// { {
// obidebug(1, "\nError creating the second column for the sequence counts when aligning"); obidebug(1, "\nError creating the second column for the sequence counts when aligning");
// return -1; return -1;
// } }
// } }
return 0; return 0;
} }
@ -267,11 +267,11 @@ static int print_alignment_result(Obiview_p output_view,
OBIDMS_column_p seq2_column, OBIDMS_column_p seq2_column,
index_t seq1_idx, index_t seq1_idx,
index_t seq2_idx, index_t seq2_idx,
// bool print_count, bool print_count,
// OBIDMS_column_p count1_column, OBIDMS_column_p count1_column,
// OBIDMS_column_p count2_column, OBIDMS_column_p count2_column,
// int count1, int count1,
// int count2, int count2,
OBIDMS_column_p ali_length_column, OBIDMS_column_p ali_length_column,
int ali_length, int ali_length,
OBIDMS_column_p lcs_length_column, OBIDMS_column_p lcs_length_column,
@ -322,21 +322,21 @@ static int print_alignment_result(Obiview_p output_view,
} }
} }
// // Write the counts if needed // TODO count columns not implemented yet // Write the counts if needed
// if (print_count) if (print_count)
// { {
// if (obi_set_index_with_elt_idx_and_col_p_in_view(output_view, count1_column, line, 0, count1) < 0) if (obi_set_int_with_elt_idx_and_col_p_in_view(output_view, count1_column, line, 0, count1) < 0)
// { {
// obidebug(1, "\nError writing count1 in a column"); obidebug(1, "\nError writing count1 in a column");
// return -1; return -1;
// } }
//
// if (obi_set_index_with_elt_idx_and_col_p_in_view(output_view, count2_column, line, 0, count2) < 0) if (obi_set_int_with_elt_idx_and_col_p_in_view(output_view, count2_column, line, 0, count2) < 0)
// { {
// obidebug(1, "\nError writing count2 in a column"); obidebug(1, "\nError writing count2 in a column");
// return -1; return -1;
// } }
// } }
// Write the alignment length if it was computed // Write the alignment length if it was computed
if ((reference == ALILEN) && (normalize || !similarity_mode)) if ((reference == ALILEN) && (normalize || !similarity_mode))
@ -385,9 +385,13 @@ static int print_alignment_result(Obiview_p output_view,
**********************************************************************/ **********************************************************************/
int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char* seq_column_name, const char* seq_elt_name, int obi_lcs_align_one_column(const char* dms_name,
const char* seq_view_name,
const char* seq_column_name,
const char* seq_elt_name,
const char* id_column_name, const char* id_column_name,
const char* output_view_name, const char* output_view_comments, const char* output_view_name,
const char* output_view_comments,
bool print_seq, bool print_count, bool print_seq, bool print_count,
double threshold, bool normalize, int reference, bool similarity_mode, double threshold, bool normalize, int reference, bool similarity_mode,
int thread_count) int thread_count)
@ -396,6 +400,7 @@ int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char
index_t seq_count; index_t seq_count;
index_t id1_idx, id2_idx; index_t id1_idx, id2_idx;
index_t seq1_idx, seq2_idx; index_t seq1_idx, seq2_idx;
int count1, count2;
double score; double score;
int lcs_length; int lcs_length;
int ali_length; int ali_length;
@ -405,16 +410,18 @@ int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char
int lcs_min; int lcs_min;
index_t seq_elt_idx; index_t seq_elt_idx;
OBIDMS_p dms = NULL;
Obiview_p seq_view = NULL; Obiview_p seq_view = NULL;
Obiview_p output_view = NULL; Obiview_p output_view = NULL;
OBIDMS_column_p iseq_column = NULL; OBIDMS_column_p iseq_column = NULL;
OBIDMS_column_p i_count_column = NULL;
OBIDMS_column_p id_column = NULL; OBIDMS_column_p id_column = NULL;
OBIDMS_column_p id1_column = NULL; OBIDMS_column_p id1_column = NULL;
OBIDMS_column_p id2_column = NULL; OBIDMS_column_p id2_column = NULL;
OBIDMS_column_p seq1_column = NULL; OBIDMS_column_p seq1_column = NULL;
OBIDMS_column_p seq2_column = NULL; OBIDMS_column_p seq2_column = NULL;
//OBIDMS_column_p count1_column = NULL; OBIDMS_column_p count1_column = NULL;
//OBIDMS_column_p count2_column = NULL; OBIDMS_column_p count2_column = NULL;
OBIDMS_column_p idx1_column = NULL; OBIDMS_column_p idx1_column = NULL;
OBIDMS_column_p idx2_column = NULL; OBIDMS_column_p idx2_column = NULL;
OBIDMS_column_p lcs_length_column = NULL; OBIDMS_column_p lcs_length_column = NULL;
@ -423,6 +430,14 @@ int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char
k = 0; k = 0;
// Open DMS
dms = obi_open_dms(dms_name);
if (dms == NULL)
{
obidebug(1, "\nError opening the DMS");
return -1;
}
// Open input view // Open input view
seq_view = obi_open_view(dms, seq_view_name); seq_view = obi_open_view(dms, seq_view_name);
if (seq_view == NULL) if (seq_view == NULL)
@ -494,6 +509,17 @@ int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char
return -1; return -1;
} }
// Open the input count column
if (print_count)
{
i_count_column = obi_view_get_column(seq_view, COUNT_COLUMN);
if (i_count_column == NULL)
{
obidebug(1, "\nError getting the input COUNT column");
return -1;
}
}
// Create the output view // Create the output view
output_view = obi_new_view(dms, output_view_name, NULL, NULL, output_view_comments); output_view = obi_new_view(dms, output_view_name, NULL, NULL, output_view_comments);
if (output_view == NULL) if (output_view == NULL)
@ -521,11 +547,11 @@ int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char
seq1_column = obi_view_get_column(output_view, SEQ1_COLUMN_NAME); seq1_column = obi_view_get_column(output_view, SEQ1_COLUMN_NAME);
seq2_column = obi_view_get_column(output_view, SEQ2_COLUMN_NAME); seq2_column = obi_view_get_column(output_view, SEQ2_COLUMN_NAME);
} }
// if (print_count) // TODO count columns not implemented yet if (print_count)
// { {
// count1_column = obi_view_get_column(seq_view, COUNT1_COLUMN_NAME); count1_column = obi_view_get_column(output_view, COUNT1_COLUMN_NAME);
// count2_column = obi_view_get_column(seq_view, COUNT2_COLUMN_NAME); count2_column = obi_view_get_column(output_view, COUNT2_COLUMN_NAME);
// } }
// Build kmer tables // Build kmer tables
ktable = hash_seq_column(seq_view, iseq_column, seq_elt_idx); ktable = hash_seq_column(seq_view, iseq_column, seq_elt_idx);
@ -597,11 +623,18 @@ int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char
// Get second id idx // Get second id idx
id2_idx = obi_get_index_with_elt_idx_and_col_p_in_view(seq_view, id_column, j, 0); id2_idx = obi_get_index_with_elt_idx_and_col_p_in_view(seq_view, id_column, j, 0);
// Get counts // TODO use array for efficiency?
if (print_count)
{
count1 = obi_get_int_with_elt_idx_and_col_p_in_view(seq_view, i_count_column, i, 0);
count2 = obi_get_int_with_elt_idx_and_col_p_in_view(seq_view, i_count_column, j, 0);
}
if (print_alignment_result(output_view, k, if (print_alignment_result(output_view, k,
idx1_column, idx2_column, i, j, idx1_column, idx2_column, i, j,
id1_column, id2_column, id1_idx, id2_idx, id1_column, id2_column, id1_idx, id2_idx,
print_seq, seq1_column, seq2_column, seq1_idx, seq2_idx, print_seq, seq1_column, seq2_column, seq1_idx, seq2_idx,
//print_count, count1_column, count2_column, count1, count2, print_count, count1_column, count2_column, count1, count2,
ali_length_column, ali_length, ali_length_column, ali_length,
lcs_length_column, lcs_length, lcs_length_column, lcs_length,
score_column, score, score_column, score,
@ -625,13 +658,19 @@ int obi_lcs_align_one_column(OBIDMS_p dms, const char* seq_view_name, const char
return -1; return -1;
} }
if (obi_close_dms(dms, false) < 0)
{
obidebug(1, "\nError closing the DMS after aligning");
return -1;
}
free_kmer_tables(ktable, seq_count); free_kmer_tables(ktable, seq_count);
return 0; return 0;
} }
int obi_lcs_align_two_columns(OBIDMS_p dms, int obi_lcs_align_two_columns(const char* dms_name,
const char* seq1_view_name, const char* seq1_view_name,
const char* seq2_view_name, const char* seq2_view_name,
const char* seq1_column_name, const char* seq1_column_name,
@ -640,7 +679,8 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
const char* seq2_elt_name, const char* seq2_elt_name,
const char* id1_column_name, const char* id1_column_name,
const char* id2_column_name, const char* id2_column_name,
const char* output_view_name, const char* output_view_comments, const char* output_view_name,
const char* output_view_comments,
bool print_seq, bool print_count, bool print_seq, bool print_count,
double threshold, bool normalize, int reference, bool similarity_mode) double threshold, bool normalize, int reference, bool similarity_mode)
{ {
@ -649,6 +689,7 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
index_t seq2_count; index_t seq2_count;
index_t id1_idx, id2_idx; index_t id1_idx, id2_idx;
index_t seq1_idx, seq2_idx; index_t seq1_idx, seq2_idx;
int count1, count2;
double score; double score;
int lcs_length; int lcs_length;
int ali_length; int ali_length;
@ -660,6 +701,7 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
index_t seq2_elt_idx; index_t seq2_elt_idx;
bool same_indexer; bool same_indexer;
OBIDMS_p dms = NULL;
Obiview_p seq1_view = NULL; Obiview_p seq1_view = NULL;
Obiview_p seq2_view = NULL; Obiview_p seq2_view = NULL;
Obiview_p output_view = NULL; Obiview_p output_view = NULL;
@ -667,12 +709,14 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
OBIDMS_column_p i_seq2_column = NULL; OBIDMS_column_p i_seq2_column = NULL;
OBIDMS_column_p i_id1_column = NULL; OBIDMS_column_p i_id1_column = NULL;
OBIDMS_column_p i_id2_column = NULL; OBIDMS_column_p i_id2_column = NULL;
OBIDMS_column_p i_count1_column = NULL;
OBIDMS_column_p i_count2_column = NULL;
OBIDMS_column_p id1_column = NULL; OBIDMS_column_p id1_column = NULL;
OBIDMS_column_p id2_column = NULL; OBIDMS_column_p id2_column = NULL;
OBIDMS_column_p seq1_column = NULL; OBIDMS_column_p seq1_column = NULL;
OBIDMS_column_p seq2_column = NULL; OBIDMS_column_p seq2_column = NULL;
//OBIDMS_column_p count1_column = NULL; OBIDMS_column_p count1_column = NULL;
//OBIDMS_column_p count2_column = NULL; OBIDMS_column_p count2_column = NULL;
OBIDMS_column_p idx1_column = NULL; OBIDMS_column_p idx1_column = NULL;
OBIDMS_column_p idx2_column = NULL; OBIDMS_column_p idx2_column = NULL;
OBIDMS_column_p lcs_length_column = NULL; OBIDMS_column_p lcs_length_column = NULL;
@ -681,6 +725,14 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
k = 0; k = 0;
// Open DMS
dms = obi_open_dms(dms_name);
if (dms == NULL)
{
obidebug(1, "\nError opening the DMS to align");
return -1;
}
// Open the first input view // Open the first input view
seq1_view = obi_open_view(dms, seq1_view_name); seq1_view = obi_open_view(dms, seq1_view_name);
if (seq1_view == NULL) if (seq1_view == NULL)
@ -835,6 +887,23 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
return -1; return -1;
} }
// Open the input count columns
if (print_count)
{
i_count1_column = obi_view_get_column(seq1_view, COUNT_COLUMN);
if (i_count1_column == NULL)
{
obidebug(1, "\nError getting the first input COUNT column");
return -1;
}
i_count2_column = obi_view_get_column(seq2_view, COUNT_COLUMN);
if (i_count2_column == NULL)
{
obidebug(1, "\nError getting the second input COUNT column");
return -1;
}
}
// Create the output view // Create the output view
output_view = obi_new_view(dms, output_view_name, NULL, NULL, output_view_comments); output_view = obi_new_view(dms, output_view_name, NULL, NULL, output_view_comments);
if (output_view == NULL) if (output_view == NULL)
@ -862,11 +931,11 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
seq1_column = obi_view_get_column(output_view, SEQ1_COLUMN_NAME); seq1_column = obi_view_get_column(output_view, SEQ1_COLUMN_NAME);
seq2_column = obi_view_get_column(output_view, SEQ2_COLUMN_NAME); seq2_column = obi_view_get_column(output_view, SEQ2_COLUMN_NAME);
} }
// if (print_count) // TODO count columns not implemented yet if (print_count)
// { {
// count1_column = obi_view_get_column(seq_view, COUNT1_COLUMN_NAME); count1_column = obi_view_get_column(output_view, COUNT1_COLUMN_NAME);
// count2_column = obi_view_get_column(seq_view, COUNT2_COLUMN_NAME); count2_column = obi_view_get_column(output_view, COUNT2_COLUMN_NAME);
// } }
// Check if the sequence columns share the same indexer (allows for quick checking of sequence equality) // Check if the sequence columns share the same indexer (allows for quick checking of sequence equality)
if (strcmp((i_seq1_column->header)->indexer_name, (i_seq2_column->header)->indexer_name) == 0) if (strcmp((i_seq1_column->header)->indexer_name, (i_seq2_column->header)->indexer_name) == 0)
@ -949,11 +1018,18 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
// Get second id idx // Get second id idx
id2_idx = obi_get_index_with_elt_idx_and_col_p_in_view(seq2_view, i_id2_column, j, 0); id2_idx = obi_get_index_with_elt_idx_and_col_p_in_view(seq2_view, i_id2_column, j, 0);
// Get counts // TODO use array for efficiency?
if (print_count)
{
count1 = obi_get_int_with_elt_idx_and_col_p_in_view(seq1_view, i_count1_column, i, 0);
count2 = obi_get_int_with_elt_idx_and_col_p_in_view(seq2_view, i_count2_column, j, 0);
}
if (print_alignment_result(output_view, k, if (print_alignment_result(output_view, k,
idx1_column, idx2_column, i, j, idx1_column, idx2_column, i, j,
id1_column, id2_column, id1_idx, id2_idx, id1_column, id2_column, id1_idx, id2_idx,
print_seq, seq1_column, seq2_column, seq1_idx, seq2_idx, print_seq, seq1_column, seq2_column, seq1_idx, seq2_idx,
//print_count, count1_column, count2_column, count1, count2, print_count, count1_column, count2_column, count1, count2,
ali_length_column, ali_length, ali_length_column, ali_length,
lcs_length_column, lcs_length, lcs_length_column, lcs_length,
score_column, score, score_column, score,
@ -986,6 +1062,12 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
return -1; return -1;
} }
if (obi_close_dms(dms, false) < 0)
{
obidebug(1, "\nError closing the DMS after aligning");
return -1;
}
free_kmer_tables(ktable, seq1_count + seq2_count); free_kmer_tables(ktable, seq1_count + seq2_count);
return 0; return 0;

26
src/obi_align.h → src/obi_lcs.h
View File

@ -3,15 +3,15 @@
****************************************************************************/ ****************************************************************************/
/** /**
* @file obi_align.h * @file obi_lcs.h
* @author Celine Mercier (celine.mercier@metabarcoding.org) * @author Celine Mercier (celine.mercier@metabarcoding.org)
* @date May 11th 2016 * @date May 11th 2016
* @brief Header file for the functions handling the LCS alignment of DNA sequences. * @brief Header file for the functions handling the LCS alignment of DNA sequences.
*/ */
#ifndef OBI_ALIGN_H_ #ifndef OBI_LCS_H_
#define OBI_ALIGN_H_ #define OBI_LCS_H_
#include <stdlib.h> #include <stdlib.h>
@ -59,7 +59,7 @@
* *
* Note: The columns where the results are written are automatically named and created. * Note: The columns where the results are written are automatically named and created.
* *
* @param dms A pointer on an OBIDMS. * @param dms_name The path of the DMS.
* @param seq_view_name The name of the view where the column to align is. * @param seq_view_name The name of the view where the column to align is.
* @param seq_column_name The name of the OBI_SEQ column in the input view to align. * @param seq_column_name The name of the OBI_SEQ column in the input view to align.
* If "" (empty string), and the input view is of type NUC_SEQS_VIEW, the associated "NUC_SEQ" column is aligned. * If "" (empty string), and the input view is of type NUC_SEQS_VIEW, the associated "NUC_SEQ" column is aligned.
@ -87,10 +87,13 @@
* @since May 2016 * @since May 2016
* @author Celine Mercier (celine.mercier@metabarcoding.org) * @author Celine Mercier (celine.mercier@metabarcoding.org)
*/ */
int obi_lcs_align_one_column(OBIDMS_p dms, int obi_lcs_align_one_column(const char* dms_name,
const char* seq_view_name, const char* seq_column_name, const char* seq_elt_name, const char* seq_view_name,
const char* seq_column_name,
const char* seq_elt_name,
const char* id_column_name, const char* id_column_name,
const char* output_view_name, const char* output_view_comments, const char* output_view_name,
const char* output_view_comments,
bool print_seq, bool print_count, bool print_seq, bool print_count,
double threshold, bool normalize, int reference, bool similarity_mode, double threshold, bool normalize, int reference, bool similarity_mode,
int thread_count); int thread_count);
@ -103,7 +106,7 @@ int obi_lcs_align_one_column(OBIDMS_p dms,
* *
* Note: The columns where the results are written are automatically named and created. * Note: The columns where the results are written are automatically named and created.
* *
* @param dms A pointer on an OBIDMS. * @param dms_name The path of the DMS.
* @param seq1_view_name The name of the view where the first column to align is. * @param seq1_view_name The name of the view where the first column to align is.
* @param seq2_view_name The name of the view where the second column to align is ("" if it is the same view as the first one). * @param seq2_view_name The name of the view where the second column to align is ("" if it is the same view as the first one).
* @param seq1_column_name The name of the first OBI_SEQ column in the input view to align. * @param seq1_column_name The name of the first OBI_SEQ column in the input view to align.
@ -138,7 +141,7 @@ int obi_lcs_align_one_column(OBIDMS_p dms,
* @since December 2016 * @since December 2016
* @author Celine Mercier (celine.mercier@metabarcoding.org) * @author Celine Mercier (celine.mercier@metabarcoding.org)
*/ */
int obi_lcs_align_two_columns(OBIDMS_p dms, int obi_lcs_align_two_columns(const char* dms_name,
const char* seq1_view_name, const char* seq1_view_name,
const char* seq2_view_name, const char* seq2_view_name,
const char* seq1_column_name, const char* seq1_column_name,
@ -147,10 +150,11 @@ int obi_lcs_align_two_columns(OBIDMS_p dms,
const char* seq2_elt_name, const char* seq2_elt_name,
const char* id1_column_name, const char* id1_column_name,
const char* id2_column_name, const char* id2_column_name,
const char* output_view_name, const char* output_view_comments, const char* output_view_name,
const char* output_view_comments,
bool print_seq, bool print_count, bool print_seq, bool print_count,
double threshold, bool normalize, int reference, bool similarity_mode); double threshold, bool normalize, int reference, bool similarity_mode);
#endif /* OBI_ALIGN_H_ */ #endif /* OBI_LCS_H_ */