TABLE III:
Alignment percentages and weighted GI/BI scores for different segment lengths, of the generated sequences with extensively trained model using different proportion of public data, against the original dataset as reference. Notice that for PKMT, the public pangenome graph will change with the public data chosen changed.
| Alignment Segment | 1k | 5k | 20k | 50k | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 20% as public | Align % | GI | BI | Align % | GI | BI | Align % | GI | BI | Align % | GI | BI |
| GSNT | 81.36 | 0.8720 | 0.9956 | 56.58 | 0.8941 | 0.9912 | 17.65 | 0.8932 | 0.9901 | 5.13 | 0.8625 | 0.9912 |
| PKMT | 63.36 | 0.9848 | 0.9978 | 63.44 | 0.9016 | 0.9969 | 61.06 | 0.9045 | 0.9952 | 55.55 | 0.9014 | 0.9948 |
| 50% as public | Align % | GI | BI | Align % | GI | BI | Align % | GI | BI | Align % | GI | BI |
| GSNT | 83.23 | 0.8731 | 0.9944 | 60.12 | 0.8966 | 0.9923 | 20.26 | 0.8941 | 0.9900 | 6.75 | 0.8654 | 0.9903 |
| PKMT | 81.03 | 0.9810 | 0.9981 | 79.53 | 0.9040 | 0.9975 | 73.00 | 0.9059 | 0.9956 | 69.90 | 0.9025 | 0.9948 |