. 2013 Oct;20(10):714–737. doi: 10.1089/cmb.2013.0084

Table 1.

Comparison of Assemblers on ECOLI-SC, a Single-Cell E. coli Dataset

Assembler^a	NGA50 # contigs^b		Longest contig	Total length	MA^c	MM^d	IND^e	Ns^f	GF (%)^g	No. genes^h
Conventional (multicell) assemblers
A5	13310	745	101584	4441145	8	11.97	0.19	0.00	90.141	3453
ABySS	68534	179	178720	4345617	5	2.71	2.66	17.07	88.268	3704
CLC	32277	503	113285	4656964	3	4.76	2.87	7.40	92.378	3768
EULER-SR	26580	429	140518	4248713	18	9.37	218.72	58.14	85.005	3419
Ray	53903	296	210612	4649552	13	2.34	0.87	0.00	91.864	3838
SOAPdenovo	16606	569	87533	4098032	7	114.38	11.08	1295.26	79.861	3038
Velvet	22648	261	132865	3501984	2	2.07	1.23	0.00	74.254	3098
Single-cell assemblers
E + V-SC	32051	344	132865	4540286	2	1.85	0.70	0.00	92.162	3793
IDBA-UD	98306	244	284464	4814043	7	2.08	0.11	0.00	95.763	4062
SPAdes 2.4	110782	274	268093	4929226	2	3.28	0.49	2.52	96.157	4060

^{^a}

Comparisons were performed with QUAST 1.2 (Gurevich et al., 2013). In each column, the best assembler by that criteria is indicated in bold.

^{^b}

Only contigs of length ≥500 bp were used.

^{^c}

MA: number of misassemblies. Misassemblies are locations on an assembled contig where the left flanking sequence aligns over 1 kb away from the right flanking sequence on the reference.

^{^d}

MM: Mismatch (substitution) error rate per 100 kb.

^{^e}

IND: number of indels per 100 kb. MM and IND are measured in aligned regions of the contigs.

^{^f}

Ns: Count of undefined bases (Ns) per 100 kb.

^{^g}

GF (%): The genome fraction is the fraction of the genome covered by the contigs. For single-cell projects, the total assembly size often exceeds the genome length due to contaminants and other reasons (see Woyke et al., 2011). The genome fraction filters out these issues.

^{^h}

The number of genes sequenced at full length is out of a list of 4324 annotated genes from www.ecogene.org for E. coli.