From 4733ef5011157767940a5056b489af99b8daae7c Mon Sep 17 00:00:00 2001 From: chhylp123 Date: Fri, 31 Jan 2025 20:27:22 -0500 Subject: [PATCH] update readme for ont --- CommandLines.cpp | 54 ++++++++++++++++++++++++++++++++++++++++++++---- CommandLines.h | 1 + README.md | 3 +++ htab.cpp | 12 ++++++++++- 4 files changed, 65 insertions(+), 5 deletions(-) diff --git a/CommandLines.cpp b/CommandLines.cpp index 08ca8ae..2af9327 100644 --- a/CommandLines.cpp +++ b/CommandLines.cpp @@ -78,6 +78,7 @@ static ko_longopt_t long_options[] = { // { "sc-n", ko_no_argument, 360}, { "chem-c", ko_required_argument, 361}, { "chem-f", ko_required_argument, 362}, + { "ul-m", ko_required_argument, 363}, // { "path-round", ko_required_argument, 348}, { 0, 0, 0 } }; @@ -186,6 +187,11 @@ void Print_H(hifiasm_opt_t* asm_opt) fprintf(stderr, " Ultra-Long-integration:\n"); fprintf(stderr, " --ul FILEs file names of Ultra-Long reads [r1.fq,r2.fq,...]\n"); + ///pending for integration + /** + fprintf(stderr, " --ul-m INT\n"); + fprintf(stderr, " hybrid assembly mode. 0: fast and memory efficent; 1: may produce better assembly with ONT R10 [%d]\n", asm_opt->ul_mod); + **/ fprintf(stderr, " --ul-rate FLOAT\n"); fprintf(stderr, " error rate of Ultra-Long reads [%.3g]\n", asm_opt->ul_error_rate); fprintf(stderr, " --ul-tip INT\n"); @@ -357,6 +363,7 @@ void init_opt(hifiasm_opt_t* asm_opt) asm_opt->is_sc = 0; asm_opt->chemical_cov = 1; asm_opt->chemical_flank = 256; + asm_opt->ul_mod = 0; } void destory_enzyme(enzyme* f) @@ -430,14 +437,43 @@ static int check_file(char* name, const char* opt) static int check_hic_reads(enzyme* f, const char* opt) { - int i; - for (i = 0; i < f->n; i++) - { + int32_t i; + for (i = 0; i < f->n; i++) { if(check_file(f->a[i], opt) == 0) return 0; } return 1; } +static int check_fq_files(enzyme* f, const char* opt, int32_t is_fq) +{ + int32_t i, ret; gzFile dfp; kseq_t *ks = NULL; + for (i = 0; i < f->n; i++) { + if(!(f->a[i])) { + fprintf(stderr, "[ERROR] input file does not exist (%s)\n", opt); + return 0; + } + + dfp = gzopen(f->a[i], "r"); + if (dfp == 0) { + fprintf(stderr, "[ERROR] Cannot find the input file: %s (%s)\n", f->a[i], opt); + return 0; + } else if(is_fq){ + ks = kseq_init(dfp); + while (((ret = kseq_read(ks)) >= 0)) { + if((ks->qual.l == 0) || (ks->qual.s == NULL)) { + fprintf(stderr, "[ERROR] %s is in fasta format rather than fastq format (%s)\n", f->a[i], opt); + fprintf(stderr, "[ERROR] set --ul-m 0 for fasta files\n"); + return 0; + } + break; + } + kseq_destroy(ks); ks = NULL; + } + gzclose(dfp); + } + return 1; +} + int check_option(hifiasm_opt_t* asm_opt) { if(asm_opt->read_file_names == NULL || asm_opt->num_reads == 0) @@ -637,7 +673,7 @@ int check_option(hifiasm_opt_t* asm_opt) return 0; } - if(asm_opt->ar != NULL && check_hic_reads(asm_opt->ar, "UL") == 0) return 0; + if(asm_opt->ar != NULL && check_fq_files(asm_opt->ar, "--ul", asm_opt->ul_mod) == 0) return 0; if(asm_opt->ar != NULL && asm_opt->ar->n == 0) { fprintf(stderr, "[ERROR] wrong UL reads (--ul)\n"); @@ -686,6 +722,11 @@ int check_option(hifiasm_opt_t* asm_opt) return 0; } + if(asm_opt->ul_mod != 0 && asm_opt->ul_mod != 1) { + fprintf(stderr, "[ERROR] must be 0 or 1 (--ul-m)\n"); + return 0; + } + if(asm_opt->telo_motif) { uint64_t k, tlen = strlen((asm_opt->telo_motif)); char c; if(tlen > 32) { @@ -944,6 +985,11 @@ int CommandLine_process(int argc, char *argv[], hifiasm_opt_t* asm_opt) asm_opt->chemical_cov = atol(opt.arg); } else if (c == 362) { asm_opt->chemical_flank = atol(opt.arg); + ///pending for integration + /** + } else if (c == 363) { + asm_opt->ul_mod = atol(opt.arg); + **/ } else if (c == 'l') { ///0: disable purge_dup; 1: purge containment; 2: purge overlap asm_opt->purge_level_primary = asm_opt->purge_level_trio = atoi(opt.arg); } diff --git a/CommandLines.h b/CommandLines.h index d634eae..b7fda9e 100644 --- a/CommandLines.h +++ b/CommandLines.h @@ -137,6 +137,7 @@ typedef struct { int64_t infor_cov, s_hap_cov, trio_cov_het_ovlp; double ul_error_rate, ul_error_rate_low, ul_error_rate_hpc; int32_t ul_ec_round; + int32_t ul_mod; uint8_t is_dbg_het_cnt; uint8_t is_low_het_ul; uint8_t is_base_trans; diff --git a/README.md b/README.md index ae89d32..d7ae7a1 100644 --- a/README.md +++ b/README.md @@ -15,6 +15,9 @@ hifiasm -o CHM13.asm -t32 -l0 CHM13-HiFi.fa.gz 2> CHM13.asm.log # Assemble heterozygous genomes with built-in duplication purging hifiasm -o HG002.asm -t32 HG002-file1.fq.gz HG002-file2.fq.gz +# Assemble genomes with ONT R10 reads rather than PacBio HiFi reads using the latest release of hifiasm (>0.21.0-r686) +hifiasm -o HG002.asm --ont -t32 HG002-ont.fq.gz + # Hi-C phasing with paired-end short reads in two FASTQ files hifiasm -o HG002.asm --h1 read1.fq.gz --h2 read2.fq.gz HG002-HiFi.fq.gz diff --git a/htab.cpp b/htab.cpp index eb94fdd..d268df4 100644 --- a/htab.cpp +++ b/htab.cpp @@ -991,8 +991,18 @@ ha_ct_t *ha_count(const hifiasm_opt_t *asm_o, int flag, int HPC, int k, int w, h opt.adaLen = (keep_adapter? asm_o->adapterLen : 0); opt.min_rcnt = (low_freq?*low_freq:-1); opt.uq = (unique_only?1:0); + + n_bs = 0; + ///pending for integration + /** + if(rs && asm_o->ar && asm_o->ul_mod) { + for (i = 0; i < asm_o->ar->n; ++i){ + h = yak_count(&opt, asm_o->ar->a[i], flag|HAF_CREATE_NEW, p0, h, flt_tab, rs, us, &n_seq); + if(h) n_bs += h->bs; + } + }**/ ///asm_opt->num_reads is the number of fastq files - for (i = n_bs = 0; i < (us?1:asm_o->num_reads); ++i){ + for (i = 0; i < (us?1:asm_o->num_reads); ++i){ h = yak_count(&opt, asm_o->read_file_names[i], flag|HAF_CREATE_NEW, p0, h, flt_tab, rs, us, &n_seq); if(h) n_bs += h->bs; }