Data Sets | # of aTIS sites | Data Set Size | MSE (All) | MSE (aTIS) | MSE (non-aTIS) | % Correct | False Positives | False Negatives |
---|
Training Set
| 41 | 82 | 0.1165 | 0.1109 | 0.1221 | 82.9 | 7 | 7 |
Independent Set
| 4 | 8 | 0.0969 | 0.1436 | 0.0502 | 87.5 | 0 | 1 |
Full Set
| 45 | 7627 | 0.0471 | 0.1138 | 0.0467 | 93.4 | 495 | 8 |
- Results of the ANN on the training, testing, and full data sets are shown here. For the purpose of identifying the aTISs, the 45 validated alternative start sites were combined with 45 non-alternative start sites randomly selected from the same sequences, resulting in a set of 90 sites. 82 of the 90 sites are used for the 'Training' Set, and the remaining 8 are used as an 'independent testing' set. Further all 7627 potential start sites from the full set of 45 sequences containing alternative start sites were compiled and tested by the ANN. The full set represents all potential start sites present in the 45 positive set of mRNAs, and tests the ability of the ANN to define which sites are likely to act as an alternative initiation start site.