An example of the temporal rule mining process. a-) Schematic representation of the sliding windows approach on a sample set of sequences binding to MHC class I allele HLA A*0201. The windows are shifted with the given s and w values until all sequences are covered. The resulting rules are then filtered by their support. This process is repeated for all s values from -8 to -1 and 1 to 8. b-) Representation of the L→V rule captured with the parameters s = 6 and w = 3. For HLA A*0201, Leucine in 2nd position and Valine around 9th position is a well-known binding motif . Although this motif is present in 6 of the 7 sequences (support of 0.86), it is unlikely to be captured by position specific methods due to length variance and positional shifts.