option to skip k-mer filtering

CommandLines.cpp

@@ -29,7 +29,7 @@ void Print_H(hifiasm_opt_t* asm_opt)
     fprintf(stderr, "    -k INT        k-mer length (must be <64) [%d]\n", asm_opt->k_mer_length);
 	fprintf(stderr, "    -w INT        minimizer window size [%d]\n", asm_opt->mz_win);
 	fprintf(stderr, "    -f INT        number of bits for bloom filter [%d]\n", asm_opt->bf_shift);
-	fprintf(stderr, "    -D FLOAT      drop k-mers occuring more than FLOAT*coverage times [%.1f]\n", asm_opt->high_factor);
+	fprintf(stderr, "    -D FLOAT      drop k-mers occuring >FLOAT*coverage times [%.1f]\n", asm_opt->high_factor);
     ///fprintf(stderr, "    -l            load all overlaps from disk, can avoid overlap calculation [%d]\n", asm_opt->load_index_from_disk);
     ///fprintf(stderr, "    -i            ignore saved overlaps in *.ovlp*.bin files\n");
     fprintf(stderr, "    -i            ignore saved overlaps in *.ovlp* files\n");
@@ -69,6 +69,7 @@ void init_opt(hifiasm_opt_t* asm_opt)
 	asm_opt->bf_shift = 37;
 	asm_opt->high_factor = 7.0f;
 	asm_opt->no_HPC = 0;
+	asm_opt->no_kmer_flt = 0;
     asm_opt->k_mer_min_freq = 3;
     asm_opt->k_mer_max_freq = 66;
     asm_opt->load_index_from_disk = 1;
@@ -300,7 +301,7 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
 
     int c;
 
-    while ((c = ketopt(&opt, argc, argv, 1, "hvt:o:k:lw:m:n:r:a:b:z:x:y:p:c:d:M:P:if:D:", 0)) >= 0) {
+    while ((c = ketopt(&opt, argc, argv, 1, "hvt:o:k:lw:m:n:r:a:b:z:x:y:p:c:d:M:P:if:D:F", 0)) >= 0) {
         if (c == 'h')
         {
             Print_H(asm_opt);
@@ -320,6 +321,7 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt)
         else if (c == 'l') asm_opt->load_index_from_disk = 1; 
         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 == 'a') asm_opt->clean_round = atoi(opt.arg); 
         else if (c == 'z') asm_opt->adapterLen = atoi(opt.arg);
         else if (c == 'b') asm_opt->required_read_name = opt.arg;

CommandLines.h

@@ -19,6 +19,7 @@ typedef struct {
 	int bf_shift;
 	float high_factor;
 	int no_HPC;
+	int no_kmer_flt;
     int k_mer_min_freq;
     int k_mer_max_freq;
     int load_index_from_disk;

htab.cpp

@@ -764,7 +764,7 @@ int ha_ft_isflt(const void *hh, uint64_t y)
 
 void ha_ft_destroy(void *h)
 {
-	yak_ft_destroy((yak_ft_t*)h);
+	if (h) yak_ft_destroy((yak_ft_t*)h);
 }
 
 /*************************
@@ -794,20 +794,36 @@ void *ha_ft_gen(const hifiasm_opt_t *asm_opt, All_reads *rs)
 ha_pt_t *ha_pt_gen(const hifiasm_opt_t *asm_opt, const void *flt_tab, int read_from_store, All_reads *rs)
 {
 	int64_t cnt[YAK_N_COUNTS], tot_cnt;
-	int peak_hom, peak_het, i, extra_flag = read_from_store? HAF_RS_READ : HAF_RS_WRITE_SEQ;
+	int peak_hom, peak_het, i, extra_flag1, extra_flag2;
 	ha_ct_t *ct;
 	ha_pt_t *pt;
-	ct = ha_count(asm_opt, HAF_COUNT_EXACT|extra_flag, NULL, flt_tab, rs);
+	if (read_from_store) {
+		extra_flag1 = extra_flag2 = HAF_RS_READ;
+	} else if (rs->total_reads == 0) {
+		extra_flag1 = HAF_RS_WRITE_LEN;
+		extra_flag2 = HAF_RS_WRITE_SEQ;
+	} else {
+		extra_flag1 = HAF_RS_WRITE_SEQ;
+		extra_flag2 = HAF_RS_READ;
+	}
+	ct = ha_count(asm_opt, HAF_COUNT_EXACT|extra_flag1, NULL, flt_tab, rs);
 	fprintf(stderr, "[M::%s::%.3f*%.2f] ==> counted %ld distinct minimizer k-mers\n", __func__,
 			yak_realtime(), yak_cputime() / yak_realtime(), (long)ct->tot);
 	ha_ct_hist(ct, cnt, asm_opt->thread_num);
 	fprintf(stderr, "[M::%s] count[%d] = %ld (for sanity check)\n", __func__, YAK_MAX_COUNT, (long)cnt[YAK_MAX_COUNT]);
 	peak_hom = ha_analyze_count(YAK_N_COUNTS, cnt, &peak_het);
 	if (peak_hom > 0) fprintf(stderr, "[M::%s] peak_hom: %d; peak_het: %d\n", __func__, peak_hom, peak_het);
-	ha_ct_shrink(ct, 2, YAK_MAX_COUNT - 1, asm_opt->thread_num);
+	if (flt_tab == 0) {
-	for (i = 2, tot_cnt = 0; i <= YAK_MAX_COUNT - 1; ++i) tot_cnt += cnt[i] * i;
+		int cutoff = (int)(peak_hom * asm_opt->high_factor);
+		if (cutoff > YAK_MAX_COUNT - 1) cutoff = YAK_MAX_COUNT - 1;
+		ha_ct_shrink(ct, 2, cutoff, asm_opt->thread_num);
+		for (i = 2, tot_cnt = 0; i <= cutoff; ++i) tot_cnt += cnt[i] * i;
+	} else {
+		ha_ct_shrink(ct, 2, YAK_MAX_COUNT - 1, asm_opt->thread_num);
+		for (i = 2, tot_cnt = 0; i <= YAK_MAX_COUNT - 1; ++i) tot_cnt += cnt[i] * i;
+	}
 	pt = ha_pt_gen(ct, asm_opt->thread_num);
-	ha_count(asm_opt, HAF_COUNT_EXACT|HAF_RS_READ, pt, flt_tab, rs);
+	ha_count(asm_opt, HAF_COUNT_EXACT|extra_flag2, pt, flt_tab, rs);
 	assert((uint64_t)tot_cnt == pt->tot_pos);
 	//ha_pt_sort(pt, asm_opt->thread_num);
 	fprintf(stderr, "[M::%s::%.3f*%.2f] ==> indexed %ld positions\n", __func__,

main.cpp

@@ -12,7 +12,8 @@ int main(int argc, char *argv[])
     init_opt(&asm_opt);
     if (!CommandLine_process(argc, argv, &asm_opt)) return 1;
 	yak_reset_realtime();
-	ha_flt_tab = ha_ft_gen(&asm_opt, &R_INF);
+	if (!asm_opt.no_kmer_flt)
+		ha_flt_tab = ha_ft_gen(&asm_opt, &R_INF);
     Correct_Reads(asm_opt.number_of_round);
 	ha_ft_destroy(ha_flt_tab);
 	destory_All_reads(&R_INF);