Initial commit

read_me.md

 ### FilterDCA 
 ### interpretable supervised contact prediction using inter-domain coevolution
 
-FilterDCA used 2 features to compute a probability of being a contact for a cople (i,j) in domain1 and domain2.
-The first feauture is the result of the method plmDCA
-The second one is a pattern score wich can be computed using the script and the maps given.
+FilterDCA used 2 features to compute a probability of being a contact for a couple (i,j) in domain1 and domain2.
+The first feature is the result of the method plmDCA.
+The second one is a pattern score which is computed by apply severals maps on the dca score matrix and keeping the best correlation.
 
 To use the script you need:
 - the result of plmDCA for the join-MSA of the 2 domains ;
@@ -13,7 +13,7 @@ To use the script you need:
 
 In the 2 folders you can find:
 - the 6 maps (3 corresponding to helix-helix contact, and 3 for strand-strand contacts) for each of the possibles size (5, 13, 21, 37, 45 or 69)
-- the classifier, and the 'min' 'max' and value to normalise the correlation/ pattern score
+- the classifier and the 'min'/'max' values to normalise the correlation/ pattern score
 
 You can then produice :
 - the pattern score: the best correlation score matrix