Table 3.
Statistics for haplotype reconstruction with tuned parameters
| Dataset | Software | Parameter values | Recalla, b | Precisionb, c | RMSDd |
|---|---|---|---|---|---|
| 50/50 | QuRe | 1E-25 0.00035 100 | 1 | 1.00 | 9.76 |
| ViQuaS | 20 0.7 | 1 | 0.67 | 23.72 | |
| CliqueSNV | t 10 tf 0.01 | 1 | 1.00 | 7.20 | |
| 70/30 | QuRe | 1E-25 0,00035 100 | 1 | 1.00 | 0.55 |
| ViQuaS | 35 0.7 | 1 | 0.67 | 14.49 | |
| CliqueSNV | t 10 tf 0.01 | 1 | 1.00 | 6.59 | |
| 90/10 | QuRe | 1E-25 0.00035 100 | 0.5 | 1.00 | 10.00 |
| ViQuaS | 5 0.7 | 1 | 0.67 | 2.78 | |
| CliqueSNV | t 10 tf 0.01 | 1 | 1.00 | 2.00 |
aRecall was calculated as the true positive haplotypes/expected number of haplotypes.
bOnly one mutation (with regard to the closest variant) was allowed for a reconstructed haplotype to be consider as a true positive haplotype.
cPrecision was calculated as the true positive haplotypes/total number of haplotypes reconstructed.
dRMSD is the root mean square deviation of the frequency estimations of the two expected haplotypes.