Evaluation of algorithms used for cross-species proteome characterisation

Area of Science:

• Proteomics
• Bioinformatics
• Computational Biology

Background:

• Protein identification from 2D electrophoresis gels is crucial for microbial proteome studies.
• Database search algorithms rely on N-/C-terminal microsequence, amino acid composition, and peptide-mass fingerprinting.
• Assessing algorithm effectiveness is vital for accurate protein characterization.

Purpose of the Study:

• To evaluate the effectiveness of nine web-accessible database search algorithms and custom software (COMBINED) for protein identification.
• To compare algorithm performance across species using ribosomal proteins and amino acyl tRNA synthetases with varying sequence identities.
• To predict the utility of experimental data for non-sequence-dependent software in proteome analysis.

Main Methods:

• Utilized Mycoplasma genitalium and Haemophilus influenzae, M. genitalium, M. jannaschii genome data.
• Selected 54 ribosomal proteins and 72 amino acyl tRNA synthetases for analysis.
• Statistically compared algorithm success against published sequence identity (22.7-100%).

Main Results:

• Examined the ability of analytical techniques and database query programs to detect identity at the functional group level.
• Assessed protein characterization for proteins with low homology at the gene/protein sequence level.
• Predicted the utility of experimentally acquired data for non-sequence-dependent software in proteome analysis.

Conclusions:

• 'Sequence tagging' of peptide fingerprints requires 11-20 amino acid residues for broad use in cross-species protein characterization.
• Theoretical data manipulations predict the utility of experimental data for proteome analysis software.
• Algorithm effectiveness varies, highlighting the need for robust methods in microbial proteome studies.