Generation of efficient fingerprint for GFP-like fold and computational identification of potential GFP-like homologs

There is a considerable interest in the detection of GFP-like proteins due to their structural stability and functional usefulness. GFP-like proteins share highly conserved beta-barrel fold with 11 beta-strands. However, their low sequence identity hampers efficient identification of their homologous proteins from database. In this study, an attempt was made to generate a fingerprint for efficient detection of GFP-like proteins. Overlapped conserved residues (OCR)-based approach has been used to design a protein fingerprint based on sequentially and structurally conserved residues in secondary structures to detect homologous proteins very efficiently. Therefore, a fingerprint for GFP-like fold was designed using the OCR approach. However, its specificity was too low to be used for the identification of novel proteins. The conserved residues of loop regions were added and optimized to improve its specificity without losing its high sensitivity. The optimized fingerprint was employed to scan NR database. A total of 20 hypothetical proteins were detected, among which nine were validated as potential GFP-like homologs.