Receiver operating characteristic (ROC) curves and the corresponding area under the curve (AUC) for the compared models were computed to simulate predictive accuracy.
To compare model performance under different scenarios we generated allele-specific read counts for both RNA and DNA controls from a Poisson distribution (see Additional file 1).
Lastly, ROC curves were calculated for each model along with an area under each respective curve in order to compare models for accuracy in predicting HCH.
While there is much less prior sensitivity under the new model compared to those observed by Oaks et al. [ 7], there is still an effect when comparing the results of the empirical data analyzed under a diffuse (M D P P ) and informative M DPP inform divergence-time prior.
In the validation cohort (SWS) areas under the curve were similar to those in the development cohort, but the earliest gestational age at which there was evidence of an improvement in the area under the curve for model 2 compared with model 1 was at 28 weeks.
Furthermore, the error in near term predictions of new cases through time was also substantially greater under the standard model compared to the behavioral models.
The underestimation of ρ F is even present to a larger extent when ρ F is estimated from data generated under the ordered model compared to the estimation under the true model.
In accordance with the reduction of the strength of crossover depicted in Figure 5, there is an increase of LD between duplicates under the WRC model compared to the SCC model.
Stranden and Gianola [ 28] have pointed out that animals that are phenotypic outliers will exhibit more extreme predictions of genetic merit under the BN model compared to the heavy-tailed models that mute the effects of the large residuals.
In all cases there are significant likelihood increases under the cF mixture model compared to the single frequency model, indicating the cF model always fits the data better for the same topology.
It also enhanced estimates of broad sense heritability on mean-basis, with an average of 72% for seed and 70% for straw yield under the superior models, compared to 62 and 55% for RCB model, respectively.
