De novo likelihood-based measures for comparing genome assemblies

BMC Research Notes

Table 3 Homo sapiens chr 14 assembly evaluation

Assembly likelihood scores for human chromosome 14 from the GAGE project [15] using a 10,000 read sample. The results are presented separately for the contigs and scaffolds and include the number of unassembled reads (singletons), the LAP scores computed on unmated reads (LAP reads) or mate-pairs (LAP mates), the N50 contig/scaffold sizes (N50), and the reference-corrected N50 contig/scaffold sizes (CN50). The best (maximum) value for each genome-measure combination is highlighted in bold. The results for the reference assembly (either complete genome or high-quality draft) is given in the row marked truth. In addition, we provide the results for a closely related strain and species. CGAL scores calculated from the long insert library were taken from the CGAL publication. The authors only provided scores for the top three assemblies (Bowtie2 could not successfully map reads to the SOAPdenovo assembly). All values, except the LAP and CGAL scores, were taken from the GAGE publication. A threshold probability of 1e-30 was used for calculating the LAP scores. The standard deviation for both the LAP’s reads and LAP’s mates scores is 0.15.

ISSN: 1756-0500