diff --git a/Overlaps.cpp b/Overlaps.cpp index 2acb850..cb4782f 100644 --- a/Overlaps.cpp +++ b/Overlaps.cpp @@ -26123,7 +26123,6 @@ ma_sub_t **coverage_cut_ptr, int debug_g) memset(R_INF.trio_flag, AMBIGU, R_INF.total_reads*sizeof(uint8_t)); } - ///print_binned_reads(sources, n_read, coverage_cut); clean_weak_ma_hit_t(sources, reverse_sources, n_read); @@ -26136,8 +26135,6 @@ ma_sub_t **coverage_cut_ptr, int debug_g) ma_hit_flt(sources, n_read, coverage_cut, max_hang_length, mini_overlap_length); ///fix_binned_reads(sources, n_read, coverage_cut); - - ///just need to deal with trio here ma_hit_contained_advance(sources, n_read, coverage_cut, ruIndex, max_hang_length, mini_overlap_length); @@ -26145,7 +26142,6 @@ ma_sub_t **coverage_cut_ptr, int debug_g) asg_arc_del_trans(sg, gap_fuzz); asm_opt.coverage = get_coverage(sources, coverage_cut, n_read); - if(VERBOSE >= 1) { char* unlean_name = (char*)malloc(strlen(output_file_name)+25); @@ -26155,10 +26151,9 @@ ma_sub_t **coverage_cut_ptr, int debug_g) } asg_cut_tip(sg, asm_opt.max_short_tip); - ///drop_inexact_edegs_at_bubbles(sg, bubble_dist); - + if(clean_round > 0) { double cut_step; @@ -26174,6 +26169,7 @@ ma_sub_t **coverage_cut_ptr, int debug_g) int i = 0; for (i = 0; i < clean_round; i++, drop_ratio += cut_step) { + if(drop_ratio > max_ovlp_drop_ratio) { drop_ratio = max_ovlp_drop_ratio; @@ -26282,7 +26278,7 @@ ma_sub_t **coverage_cut_ptr, int debug_g) debug_ma_hit_t(reverse_sources, coverage_cut, n_read, max_hang_length, mini_overlap_length); **/ - + ///note: don't apply asg_arc_del_too_short_overlaps() after this function!!!! rescue_contained_reads_aggressive(NULL, sg, sources, coverage_cut, ruIndex, max_hang_length, mini_overlap_length, bubble_dist, 10, 1, 0, NULL, NULL); diff --git a/README.md b/README.md index 234c3d5..0908f81 100644 --- a/README.md +++ b/README.md @@ -46,15 +46,14 @@ assembly by multiple rounds of experiments with different parameters. Hifiasm is a standalone and lightweight assembler, which does not need external libraries (except zlib). For large genomes, it can generate high-quality -assembly in a few hours. Hifiasm has been tested on human, butterfly, rice and drosophila. -In particular, hifiasm is able to assemble the 26.5Gb California redwood tree in a few days. +assembly in a few hours. Hifiasm has been tested on various large and complex datasets. The results are as follows: -|Dataset|GSize|Cov|Asm options|CPU time|Wall time|RAM|[unitig][unitig]/[contig][unitig] N50[1]| +|Dataset|GSize|Cov|Asm options|CPU time|Wall time|RAM| N50| |:---------------|-----:|-----:|:---------------------|-------:|--------:|----:|----------------:| +|[\[Mouse (C57/BL6J)\]](https://www.ncbi.nlm.nih.gov/sra/?term=SRR11606870)|2.7Gb|x25|-t48 -l0|172.9h|4.8h|76G|20.6Mb| |[\[Redwood\]](https://downloads.pacbcloud.com/public/dataset/redwood2020/)|26.5Gb|x23|-k 40 -t 64 -r 2|7274h30m|141h30m|512G|1.7Mb/1.9Mb| -[1] unitig N50 is the N50 of assembly graph with haplotype information (i.e., bubbles), while the contig N50 is the N50 of haplotype collapsed assembly (i.e., without bubbles). ## Usage