RegCloser: a robust regression approach to closing genome gaps

BMC Bioinformatics

Table 2 Comparison of the five methods on closing gaps of the plateau zokor draft genome

Methods	Draft	GapCloser	GapFiller	Sealer	Phrap	RegCloser
3 paired-end libraries
Contig length	2,527,528,914	2,538,745,631	2,530,456,764	2,528,744,942	2,528,100,353	2,527,758,194
Contig number	113,835	97,906	96,181	102,260	71,066	69,093
Contig N50 (bp)	67,378	76,359	88,200	73,199	130,114	143,871
# closed gaps		15,929	17,654	11,575	42,769	44,742
BUSCO	95.1%	95.2%	95.1%	95.2%	95.2%	95.2%
Mapping rate	79.1%	79.4%	79.3%	79.2%	79.7%	79.9%
3 paired-end + 1 mate-pair libraries
Contig length	2,527,528,914	2,541,442,324	2,532,550,654	2,528,949,290	2,526,621,405	2,524,372,994
Contig number	113,835	96,437	92,755	100,611	65,445	62,766
Contig N50 (bp)	67,378	77,295	93,380	74,852	146,599	173,542
# closed gaps		17,398	21,080	13,224	48,390	51,069
BUSCO	95.1%	95.2%	95.1%	95.2%	95.2%	95.2%
Mapping rate	79.1%	79.5%	79.4%	79.2%	79.7%	80.1%

The draft genome is initially assembled with high coverage short reads, and then improved with long reads. Two rounds of gap closing are performed. In the first round, three paired-end libraries with insert sizes of 300, 500, and 800 bp are used. In the second round, a mate-pair library with a long insert size of 3000 bp is added. Mapping rate is measured by mapping the reads of an independent library to the assembly. Contig N50 assesses the contiguity of genome assemblies. BUSCO and mapping rate assess the completeness of genome assemblies. The best values of each quality metric are highlighted in bold. After both rounds, RegCloser closes the most gaps, and achieves the highest contig N50, BUSCO, and mapping rate

ISSN: 1471-2105