Table 1. Metrics for LACHESIS-based scaffolding of shotgun assemblies.

The human and mouse shotgun assemblies are based on read-pairs from short-insert and ~2.5 Kb jumping libraries, while the Drosophila shogun assembly is based solely on read-pairs from short-insert libraries⁶. The human and mouse shotgun assemblies consist of scaffolds, whereas the Drosophila shotgun assembly consists of contigs. LACHESIS places scaffolds or contigs into groups and then orders and orients them within each group. An ordering error means that a contig or scaffold’s position is out of the expected order with respect to its neighbors. An orientation error means that its orientation is not the orientation implied by its position with respect to its immediate predecessor. ‘High-quality predictions’ refers to a subset of contigs or scaffolds whose position and orientation in their ordering is deemed more certain; the threshold for high quality is chosen for convenience for each assembly.

Metric	De novo assemblies
	Human	Mouse	Drosophila
Shotgun assembly metrics
Total assembly length, including gaps (Mb)	2,739	2,370	127
Number of contigs or scaffolds	18,921	25,964	7,109
N50 contig or ungapped scaffold size (Kb)	437	224	68
Clustering
% sequence (% contigs) clustered into groups	98.2% (71.5%)	98.0% (87.8%)	81.2% (64.3%)
% clustered sequence (% contigs) mis-clustered	0.14% (1.4%)	0.24% (0.5%)	3.4% (10.5%)
Ordering
% clustered seq. (% contigs) ordered	94.4% (55.3%)	86.7% (42.7%)	82.0% (24.5%)
% ordered seq. (% contigs) w/ ordering errors	0.5% (0.8%)	0.5% (1.1%)	4.6% (5.2%)
% ordered seq. (% contigs) w/ orientation errors	1.2% (2.5%)	1.9% (4.6%)	4.1% (6.1%)
High-quality predictions
% ordered seq. (% contigs) w/ high quality	92.8% (79.0%)	93.3% (82.9%)	94.1% (88.1%)
% high-quality seq. (% contigs) w/ ordering errors	0.3% (0.4%)	0.3% (0.7%)	3.3% (3.4%)
% high-quality seq. (% contigs) w/ orientation errors	0.4% (0.5%)	0.5% (1.0%)	2.5% (2.7%)