Statistical modeling of DNase HS data to determine best fit
No. of DNase clusters | Best | |||||
|---|---|---|---|---|---|---|
| Cluster size | % valida | Normalized (observed)b | 1 statec | 2 stated | ||
| Singlet | 123,000 | 16% | 19,680 | 13,380 | 19,587 | |
| 2 | 9039 | 54% | 4881 | 11,708 | 5161 | |
| 3 | 2565 | 80% | 2078 | 6829 | 1863 | |
| 4 | 1216 | 100% | 1216 | 2987 | 1160 | |
| 5 | 674 | 100% | 674 | 1045 | 753 | |
| 6 | 348 | 100% | 348 | 305 | 424 | |
| 7 | 165 | 100% | 165 | 76 | 205 | |
| 8 | 90 | 100% | 90 | 16 | 87 | |
| >9 | 118 | 100% | 118 | 3 | 49 | |
| Χ2 | 15,115 | 170 | ||||
a Percentage validated values were derived from real-time PCR results for each cluster size
b Normalized (observed) values are the number of DNase clusters multiplied by percentage valid
c Best 1 state represents a single Poisson distribution that best fits observed data
d Best 2 state represents two classes of Poisson distributions that best fit observed data. One class assumes 8% of the DNase HS sites are eight times more hypersensitive to DNase digestion