On-line Abstract Book

Use of Artificial Intelligence Programming Language Prolog in N-terminus De Novo Sequencing of Modified Proteins Subjected to MALDI Mass Spectrometry

Matthew J. Kelly¹, Mike May¹, Shigeki Kajihara²

Correspondence address: Matthew J. Kelly, Shimadzu Research Laboratory, Wharfside, Trafford Park, Manchester, M17 1GP United Kingdom.

Keywords: Artificial Intelligence; High Throughput; MALDI; Peptide, Modified.

Novel aspect: The use of a declarative artificial intelligence programming language to de novo sequence a modified peptide.

 

After separation by SDS-PAGE or 2D PAGE, sample protein is successively reduced, S-alkykated and guanidinated, and then its Nα-amino group is coupled to a biotinyl-cysteic acid-reagent.¹ The protein is digested with trypsin and then the derivatised N-terminal fragment is specifically isolated as a sulphonic acid derivative from the tryptic digest with avidin resins. MS/MS using PSD of the sulphonic acid derivative produces a spectrum with high intensity y and z ions, suitable for processing by de novo sequencing software.

A de novo sequencing algorithm has been developed for use with derivatised peptides of the type described above. The algorithm is implemented in Prolog, which is an artificial intelligence programming language that can be programmed declaratively. The relationships between precursor mass, y and z ions can be represented as Prolog facts and rules and used in de novo sequencing. Prolog is particularly well suited to the task because it has a depth-first backtracking facility that allows all solutions for a particular set of sequencing parameters to be found.

The algorithm derives all possible ion series within a relatively large mass tolerance. An intensity-based scoring scheme is then applied to identify series candidates that are most likely to provide sequences that match the correct N-terminus sequence, and the final results reported. The de novo sequencing software can handle fixed and variable modifications of peptides.

The initial results of testing indicate that the de novo sequencing software frequently finds the correct N-terminus sequence as the top-ranked candidate. The software will assist high-throughput N-terminus sequencing of proteins.

1. M. Yamaguchi, T. Obama, H. Kuyama, E. Ando, T. Okamura, N. Ueyama, T. Nakazawa and S. Norioka, Proceeding of the 53nd ASMS Conference, San Antonio, Texas, June 5-9, A051428 (2005).