RegCloser: a robust regression approach to closing genome gaps

BMC Bioinformatics

Table 1 Comparison of the five methods on the E. coli simulation data

Methods	Draft	GapCloser	GapFiller	Sealer	Phrap	RegCloser
Contig length	4,531,075	4,642,829	4,631,201	4,542,679	4,641,796	4,641,652
Contig number	137	1	26	59	7	1
Contig N50	78,557	4,642,829	331,261	174,037	1,334,246	4,641,652
Genome fraction	97.532%	99.924%	99.687%	97.851%	99.892%	100%
# mis-assemblies	0	0	0	1	0	0
# local mis-assemblies	0	25	7	20	23	0
# mismatches	0	245	43	55	313	51
# indels	0	32	13	17	42	12
# closed gaps (# total gaps = 136)		136	111	78	130	136
# correctly closed gaps		112	103	57	108	136
# closed TRs (# total TRs = 26)		26	12	25	24	26
# correctly closed TRs		8	4	6	6	26
# incorrectly closed TRs/ # incorrectly closed gaps		18/24	8/8	19/21	18/22	0/0

The assemblies are aligned to the reference genome using QUAST 5.2.0. Genome fraction is the percentage of the reference genome covered by assembled contigs. Mis-assemblies are locations on assembled contigs where the left and right flanking sequences align over 1 kb away, or they overlap by > 1 kb, or they align on opposite strands. Local mis-assemblies are positions on contigs where the flanking sequences have a gap or overlap < 1 kbp and > 80 bp on the same strand of the reference. The best values of each quality metric are highlighted in bold. RegCloser correctly closes all the 136 gaps including the 26 tandem repeat (TR)-related gaps, and leads to a complete genome with 100% genome fraction and no mis-assemblies or local mis-assemblies. For the other four methods, the TR-related gaps account for most of the incorrectly closed gaps

ISSN: 1471-2105