Table 2.

Estimation of the node multiplicity in de Bruijn graphs (k=21) built from real Illumina data for 5 organisms (2 bacteria, 3 eukaryotes)

		10×			25×			50×
	s	node acc.	k-mer acc.		node acc.	k-mer acc.		node acc.	k-mer acc.
P. aeruginosa	0	84.31	96.47		95.02	98.72		97.47	99.21
	1	92.85	98.80		98.27	99.49		98.89	99.46
	3	93.91	98.96		98.60	99.50		99.13	99.52
	5	94.11	98.94		98.74	99.51		99.17	99.51
S. enterica	0	84.50	96.46		93.65	98.25		95.96	98.53
	1	88.81	97.18		94.98	98.39		96.44	98.60
	3	89.41	97.18		95.27	98.45		96.53	98.63
	5	89.55	97.22		95.32	98.46		96.57	98.64
C. elegans	0	68.65	93.93		80.69	96.42		87.35	97.10
	1	78.90	96.48		86.47	97.77		90.74	98.05
	3	81.02	97.21		87.16	98.01		91.27	98.24
	5	81.29	97.18		87.32	98.05		91.25	98.25
A. thaliana	0	67.84	89.10		82.20	96.16		89.91	97.05
	1	73.67	94.64		85.45	96.83		91.27	97.54
	3	73.92	95.26		85.83	97.09		91.46	97.71
	5	73.93	95.43		85.68	97.17		91.56	97.70
H. sapiens	0	75.26	92.27		83.29	94.67		88.09	95.51
	1	80.68	93.92		85.66	95.23		89.23	95.83
	3	81.33	94.56		86.12	95.50		89.52	95.95
	5	81.57	94.71		86.26	95.58		89.59	95.97

The datasets were downsampled to coverage depths of 10×,25× and 50×. For H. sapiens, the multiplicity was inferred for one million randomly sampled nodes; for all other datasets the multiplicity was inferred for all nodes. The node (resp. k-mer) accuracy refers to the percentage of nodes (resp. k-mers) in the de Bruijn graph that were assigned the correct multiplicity. The accuracy improves when using CRFs with increasing neighbourhood size s