Table 3.
Computer simulation of the experimental analysis confirms its efficacy to detect biased haplotypic input
| Model | Filter | % of dataset | HET:HOM | HET:HOM | Two-tailed |
|---|---|---|---|---|---|
| Level | Retained | Input as P(HET) | Output | Binomial p = | |
| 5 cells | 0.2–0.8 | 14.5 | 0.3 | 96:49 | 5.88E−05 |
| 14.3 | 0.275 | 90:53 | 1.24E−03 | ||
| 16.5 | 0.25 | 82:83 | ns | ||
| 10 cells | 0.2–0.8 | 1.2 | 0.3 | 11:1 | 6.35E−03 |
| 1.8 | 0.275 | 13:5 | 9.63E−02 | ||
| 1.0 | 0.25 | 4:6 | ns | ||
| 10 cells | 0.3–0.7 | 12.6 | 0.3 | 92:34 | 2.36E−07 |
| 14.6 | 0.275 | 97:49 | 8.77E−05 | ||
| 11.4 | 0.25 | 56:58 | ns |
The results of the simulation for models with 5 and 10 “cells” per “well.” Output bias of HET > HOM was tested using the binomial test. In each of the three simulations, when input was unbiased, no significant bias was exhibited in the output. In contrast, when input was biased (HET > HOM), significant bias of HET > HOM was detected in five of the six tests (p < 0.01)