PFRMAT     AL 
TARGET     T0056 
AUTHOR     7545-6133-3082 
REMARK      
METHOD   blastp was used to find homologous sequences of which a multiple   
METHOD   sequences alignment was build and used as seed for HMMer. The 
METHOD   resulting hmm was used to search against a version of pdb which 
METHOD   only contains proteins of less than 90% homology. No significant 
METHOD   hits were found during this procedure. 
METHOD   A specifically developed neural network scored all alignments on 
METHOD   basis of their secondary structure overlap. The top ten showed 
METHOD   no common fold and generally low scores. The highest score was 
METHOD   achieved by a calcium binding protein, which makes it unlikely 
METHOD   that both proteins are related. However, the third best score was  
METHOD   achieved by a DNA-topoisomerase, aprotein with extremly similar 
METHOD   function. Thus it was decided to use this protein as a parent. 
METHOD   However, as no common fold could be found amongst the top ten 
METHOD   results of the program, a blind prediction would have resulted 
METHOD   in the prediction that there is no similar fold in the database. 
METHOD   Thus the second model is for unknown fold. 
MODEL 2 
PARENT NONE 
TER 
END 
