Graph illustrating two remote homologs (1cem and 1qazA) successfully detected by the SSEA descriptor. With the predicted secondary structure from PSI-PRED, all the FSSP607 sequences were converted into the format of predicted secondary structure elements as described in the dedicated paragraph of Materials and Methods. Taking 1cem (1,4-β-D-glucan-glucanohydrolase catalytic domain; FSSP family: 126.4.1.1.2.1) (Alzari et al. 1996) as the query sequence against all other proteins in FSSP607, the remote homolog 1qazA (Alginate lyase A1-III from Sphin-gomonas Species; Chain: A; FSSP family index: 126.4.3.2.1.1) (Yoon et al. 1999) was selected as the top hit with an SSEA score of 0.7041. The identity between the two sequences was only 11%. (A) The SSEA alignment between 1cem and 1qazA. The identical secondary structure elements in the same alignment position were displayed in bold type. (B) The Cα superimposed models of 1cem (cyan) and 1qazA (purple). The superimposition was carried out by using CE algorithm (Shindyalov and Bourne 1998). With the 233 aligned residues, the root mean square distance (RMSD) for the superimposed structures is 4.3 Å, and the Z-score is 5.6, implying that they should share higher than just fold-level remote homology.