r211: added --write-paf/--write-ec/--skip-triobin

Assembly.cpp

@@ -1156,11 +1156,13 @@ int ha_assemble(void)
 	if (asm_opt.load_index_from_disk && load_all_data_from_disk(&R_INF.paf, &R_INF.reverse_paf, asm_opt.output_file_name)) {
 		ovlp_loaded = 1;
 		fprintf(stderr, "[M::%s::%.3f*%.2f] ==> loaded corrected reads and overlaps from disk\n", __func__, yak_realtime(), yak_cpu_usage());
-		trio_partition();
+		if (!(asm_opt.flag & HA_F_SKIP_TRIOBIN)) trio_partition();
+		if (asm_opt.flag & HA_F_WRITE_EC) Output_corrected_reads();
+		if (asm_opt.flag & HA_F_WRITE_PAF) Output_PAF();
 	}
 	if (!ovlp_loaded) {
 		// construct hash table for high occurrence k-mers
-		if (!asm_opt.no_kmer_flt) {
+		if (!(asm_opt.flag & HA_F_NO_KMER_FLT)) {
 			ha_flt_tab = ha_ft_gen(&asm_opt, &R_INF, &hom_cov);
 			ha_opt_update_cov(&asm_opt, hom_cov);
 		}
@@ -1175,7 +1177,7 @@ int ha_assemble(void)
 					asm_opt.num_bases, asm_opt.num_corrected_bases, asm_opt.num_recorrected_bases);
 			fprintf(stderr, "[M::%s] size of buffer: %.3fGB\n", __func__, asm_opt.mem_buf / 1073741824.0);
 		}
-		//Output_corrected_reads();
+		if (asm_opt.flag & HA_F_WRITE_EC) Output_corrected_reads();
 		fprintf(stderr, "[M::%s::%.3f*%.2f] ==> written corrected reads to disk\n", __func__, yak_realtime(), yak_cpu_usage());
 		// overlap between corrected reads
 		ha_opt_reset_to_round(&asm_opt, asm_opt.number_of_round);
@@ -1184,7 +1186,7 @@ int ha_assemble(void)
 				yak_cpu_usage(), yak_peakrss_in_gb());
 		ha_print_ovlp_stat(R_INF.paf, R_INF.reverse_paf, R_INF.total_reads);
 		ha_ft_destroy(ha_flt_tab);
-		//Output_PAF();
+		if (asm_opt.flag & HA_F_WRITE_PAF) Output_PAF();
 		trio_partition();
 	}
 	build_string_graph_without_clean(asm_opt.min_overlap_coverage, R_INF.paf, R_INF.reverse_paf, 

CommandLines.cpp

@@ -1,9 +1,9 @@
-#include "CommandLines.h"
 #include <zlib.h>
 #include <stdlib.h>
 #include <stdio.h>
-#include "ketopt.h"
 #include <sys/time.h>
+#include "CommandLines.h"
+#include "ketopt.h"
 
 #define DEFAULT_OUTPUT "hifiasm.asm"
 
@@ -12,6 +12,9 @@ hifiasm_opt_t asm_opt;
 static ko_longopt_t long_options[] = {
 	{ "version",      ko_no_argument,       300 },
 	{ "dbg-gfa",      ko_no_argument,       301 },
+	{ "write-paf",    ko_no_argument,       302 },
+	{ "write-ec",     ko_no_argument,       303 },
+	{ "skip-triobin", ko_no_argument,       304 },
 	{ 0, 0, 0 }
 };
 
@@ -58,6 +61,7 @@ void Print_H(hifiasm_opt_t* asm_opt)
 
 void init_opt(hifiasm_opt_t* asm_opt)
 {
+	asm_opt->flag = 0;
     asm_opt->coverage = -1;
     asm_opt->num_reads = 0;
     asm_opt->read_file_names = NULL;
@@ -69,8 +73,6 @@ void init_opt(hifiasm_opt_t* asm_opt)
     asm_opt->k_mer_length = 51;
 	asm_opt->mz_win = 51;
 	asm_opt->bf_shift = 37;
-	asm_opt->no_HPC = 0;
-	asm_opt->no_kmer_flt = 0;
 	asm_opt->high_factor = 5.0f;
 	asm_opt->max_n_chain = 100;
     asm_opt->k_mer_min_freq = 3;
@@ -92,7 +94,6 @@ void init_opt(hifiasm_opt_t* asm_opt)
     asm_opt->max_short_tip = 3;
     asm_opt->min_cnt = 2;
     asm_opt->mid_cnt = 5;
-	asm_opt->verbose_gfa = 0;
 }
 
 void destory_opt(hifiasm_opt_t* asm_opt)
@@ -331,7 +332,7 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
         else if (c == 'i') asm_opt->load_index_from_disk = 0; 
         else if (c == 'w') asm_opt->mz_win = atoi(opt.arg);
 		else if (c == 'D') asm_opt->high_factor = atof(opt.arg);
-		else if (c == 'F') asm_opt->no_kmer_flt = 1;
+		else if (c == 'F') asm_opt->flag |= HA_F_NO_KMER_FLT;
 		else if (c == 'N') asm_opt->max_n_chain = atoi(opt.arg);
         else if (c == 'a') asm_opt->clean_round = atoi(opt.arg); 
         else if (c == 'z') asm_opt->adapterLen = atoi(opt.arg);
@@ -345,7 +346,10 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
         else if (c == 'p') asm_opt->small_pop_bubble_size = atoll(opt.arg);
         else if (c == 'm') asm_opt->large_pop_bubble_size = atoll(opt.arg);
         else if (c == 'n') asm_opt->max_short_tip = atoll(opt.arg);
-		else if (c == 301) asm_opt->verbose_gfa = 1;
+		else if (c == 301) asm_opt->flag |= HA_F_VERBOSE_GFA;
+		else if (c == 302) asm_opt->flag |= HA_F_WRITE_PAF;
+		else if (c == 303) asm_opt->flag |= HA_F_WRITE_EC;
+		else if (c == 304) asm_opt->flag |= HA_F_SKIP_TRIOBIN;
         else if (c == ':') 
         {
 			fprintf(stderr, "[ERROR] missing option argument in \"%s\"\n", argv[opt.i - 1]);

CommandLines.h

@@ -3,11 +3,19 @@
 
 #include <pthread.h>
 
-#define HA_VERSION "0.3.0-dirty-r207"
+#define HA_VERSION "0.3.0-dirty-r211"
 
 #define VERBOSE 0
 
+#define HA_F_NO_HPC       0x1
+#define HA_F_NO_KMER_FLT  0x2
+#define HA_F_VERBOSE_GFA  0x4
+#define HA_F_WRITE_EC     0x8
+#define HA_F_WRITE_PAF    0x10
+#define HA_F_SKIP_TRIOBIN 0x20
+
 typedef struct {
+	int flag;
     int num_reads;
     char** read_file_names;
     char* output_file_name;
@@ -18,8 +26,6 @@ typedef struct {
     int k_mer_length;
 	int mz_win;
 	int bf_shift;
-	int no_HPC;
-	int no_kmer_flt;
 	float high_factor; // coverage cutoff set to high_factor*hom_cov
 	int max_n_chain; // fall-back max number of chains to consider
     int k_mer_min_freq;
@@ -37,7 +43,6 @@ typedef struct {
     int max_short_tip;
     int min_cnt;
     int mid_cnt;
-	int verbose_gfa;
 
     float max_hang_rate;
     float min_drop_rate;

Hash_Table.cpp

@@ -3,9 +3,6 @@
 #include <stdint.h>
 #include <pthread.h>
 #include "Hash_Table.h"
-#include "Process_Read.h"
-#include "Correct.h"
-#include "CommandLines.h"
 #include "ksort.h"
 pthread_mutex_t output_mutex;
 

Makefile

@@ -39,7 +39,7 @@ CommandLines.o: CommandLines.h ketopt.h
 Correct.o: Correct.h Hash_Table.h htab.h Process_Read.h Overlaps.h kvec.h
 Correct.o: kdq.h CommandLines.h Levenshtein_distance.h POA.h Assembly.h
 Hash_Table.o: Hash_Table.h htab.h Process_Read.h Overlaps.h kvec.h kdq.h
-Hash_Table.o: CommandLines.h Correct.h Levenshtein_distance.h POA.h ksort.h
+Hash_Table.o: CommandLines.h ksort.h
 Levenshtein_distance.o: Levenshtein_distance.h
 Output.o: Output.h CommandLines.h
 Overlaps.o: Overlaps.h kvec.h kdq.h ksort.h Process_Read.h CommandLines.h
@@ -50,7 +50,7 @@ POA.o: CommandLines.h Correct.h Levenshtein_distance.h
 Process_Read.o: Process_Read.h Overlaps.h kvec.h kdq.h CommandLines.h
 Purge_Dups.o: ksort.h Purge_Dups.h kvec.h kdq.h Overlaps.h Hash_Table.h
 Purge_Dups.o: htab.h Process_Read.h CommandLines.h Correct.h
-Purge_Dups.o: Levenshtein_distance.h POA.h
+Purge_Dups.o: Levenshtein_distance.h POA.h kthread.h
 Trio.o: khashl.h kthread.h Process_Read.h Overlaps.h kvec.h kdq.h
 Trio.o: CommandLines.h htab.h
 anchor.o: htab.h Process_Read.h Overlaps.h kvec.h kdq.h CommandLines.h

Overlaps.cpp

@@ -26466,7 +26466,7 @@ ma_sub_t **coverage_cut_ptr, int debug_g)
 
     asg_arc_del_simple_circle_untig(sources, coverage_cut, sg, 100, 0);
 
-    if (asm_opt.verbose_gfa)
+    if (asm_opt.flag & HA_F_VERBOSE_GFA)
     {
         /*******************************for debug***************************************/
         write_debug_graph(sg, sources, coverage_cut, output_file_name, n_read, reverse_sources, ruIndex);
@@ -26545,7 +26545,7 @@ long long bubble_dist, int read_graph, int write)
     ///actually min_thres = asm_opt.max_short_tip + 1 there are asm_opt.max_short_tip reads
     min_thres = asm_opt.max_short_tip + 1;
 
-    if (asm_opt.verbose_gfa)
+    if (asm_opt.flag & HA_F_VERBOSE_GFA)
     {
         if(load_debug_graph(&sg, &sources, &coverage_cut, output_file_name, &reverse_sources, &ruIndex))
         {

anchor.cpp

@@ -62,7 +62,7 @@ void ha_get_new_candidates(ha_abuf_t *ab, int64_t rid, UC_Read *ucr, overlap_reg
 	ab->mz.n = 0, ab->n_a = 0;
 
 	// get the list of anchors
-	ha_sketch(ucr->seq, ucr->length, asm_opt.mz_win, asm_opt.k_mer_length, 0, !asm_opt.no_HPC, &ab->mz, ha_flt_tab);
+	ha_sketch(ucr->seq, ucr->length, asm_opt.mz_win, asm_opt.k_mer_length, 0, !(asm_opt.flag & HA_F_NO_HPC), &ab->mz, ha_flt_tab);
 	if (ab->mz.m > ab->old_mz_m) {
 		ab->old_mz_m = ab->mz.m;
 		REALLOC(ab->seed, ab->old_mz_m);

htab.cpp

@@ -721,7 +721,7 @@ ha_ct_t *ha_count(const hifiasm_opt_t *asm_opt, int flag, ha_pt_t *p0, const voi
 	}
 	yak_copt_init(&opt);
 	opt.k = asm_opt->k_mer_length;
-	opt.is_HPC = !asm_opt->no_HPC;
+	opt.is_HPC = !(asm_opt->flag&HA_F_NO_HPC);
 	opt.w = flag & HAF_COUNT_ALL? 1 : asm_opt->mz_win;
 	opt.bf_shift = flag & HAF_COUNT_EXACT? 0 : asm_opt->bf_shift;
 	opt.n_thread = asm_opt->thread_num;