For given probability of false alarm, the phase threshold can be determined using (5) and the probability of detection can be computed using (6).
Due to the different perspective, the family cycle is a bit different: in the demographic typology, the young phase threshold is 35 years.
On the other hand, if one uses a constant magnitude-phase threshold pair arbitrarily, the result would lead to either too many false alarms or too few detected targets, whether or not takes into account the magnitude dependence of the phase PDF.
We therefore reduced the Phase threshold to 0.60 to maximise recovery of haplotypes.
Notably, lowering the PHASE threshold to 0.60 often reduces the number of unresolved genotypes with little or no increase in false positives (Table 3).
Lowering the PHASE threshold to 0.60 led to reductions in the number of unresolved genotypes, but also in the proportion of correctly reconstructed haplotype pairs (60- NLCP range: 0-7, mean: 2.34, proportion correct: 51.2%).
Garrick et al. [ 36], suggest that lowering the Phase threshold to 0.60 reduces the number of unresolved haplotypes with little or no increase in the number of false positives.
For those datasets in which false positives occurred, error rates were as high as 12.5% (Uc3), but all others were <10% irrespective of PHASE threshold (Sim06, up to 8%; Sim11, 8.3%; Sim33, 4.9%; UcEF-1α, 2.4%).
Indeed, it may not be necessary to experimentally phase all unresolved genotypes given that biases in the four population parameter estimates examined here were always quite low when unresolved genotypes accounted for ≤ 2% of the total dataset under the 0.90 PHASE threshold.
With the exception of HO (Additional file 5, G-H), the overall strength of each relationship increased considerably as the PHASE threshold was decreased from 0.90 to 0.60, indicating that inclusion of 'marginal' haplotype pairs in the NLCP category (i.e., those with confidence probability scores of 0.61-0.89) mostly contributes noise.
