Please refer to the link: Bayesian Scoring. The number of clues basically tells how many number of weights were use to get the final updated weight for a function. Currently the value can range up to 10 if the protein structural coordinates are available and all the features are used for query. For sequence only prediction, the value can be 8 maximum. If you turn DIP=OFF, the maximum value is decreased by 1. If you turn both DIP=OFF and PROLINKS=OFF, the maximum value is decreased by 2. For a prediction that is done from only few clues is not a confident prediction. Therefore if is advisable to keep all the feature extractors on.
The description of the clues in the Master Table
| Clue 1 | Frequency of the ontolgies obtained from Blast hits |
| Clue 2 | Score for the ontology from Blast Evalues. The best evalue available for the ontology is taken (only 4 digits after decimal is shown). |
| Clue 3 | Frequency of ontologies from 3D motifs |
| Clue 4 | Score of ontologies from 3D motifs based on conservation. It is the average of scores from the motifs associated with the ontology. |
| Clue 5 | Score of ontologies from 3D folds. The best Z_score available for the function is taken. |
| Clue 6 | Frequency of ontologies from 3D folds |
| Clue 7 | Frequency of ontologies from DIP search |
| Clue 8 | Score of ontologies from PROSITE search based on conservation. It is the average of scores from the motifs associated with the ontology. |
| Clue 9 | Frequency of ontologies from PROSITE search |
| Clue 10 | Frequency of ontologies from PROLINKS search |