T3_MM: a Markov model effectively classifies bacterial type III secretion signals.

Area of Science:

• Microbiology
• Bioinformatics
• Computational Biology

Background:

• Type III Secretion Systems (T3SSs) are crucial for gram-negative bacteria-host interactions, translocating effector proteins into host cells.
• The precise mechanisms of T3SSs and the characteristics of their effector proteins require further elucidation.
• This research investigates the N-terminal amino acid composition of type III secretion (T3S) signal sequences to identify distinguishing features.

Purpose of the Study:

• To quantitatively model and distinguish T3S effector proteins.
• To analyze amino acid composition constraints within the N-terminal 100 amino acids of T3S and non-T3S proteins.
• To develop a statistical model for predicting T3S proteins.

Main Methods:

• Comparison of amino acid composition (Aac) probability profiles between N-terminal sequences of T3S and non-T3S proteins.
• Development of a Markov model (T3_MM) to calculate the likelihood ratio of a sequence being T3S or non-T3S.
• Statistical analysis of Aac conditional probabilities and residue constraints.

Main Results:

• Amino acid composition profiles of T3S and non-T3S proteins were found to be significantly different.
• The T3_MM model successfully distinguished known T3S and non-T3S proteins, approximating distinct normal distributions.
• A 5-fold cross-validation demonstrated T3_MM's high performance: 83.9% sensitivity and 90.3% specificity.
• T3_MM proved more robust, accurate, simple, and statistically quantitative than existing T3S prediction models.

Conclusions:

• The N-terminal amino acid composition provides a strong basis for distinguishing T3S effector proteins.
• The T3_MM model offers an effective and statistically sound method for T3S protein prediction.
• The findings highlight the constraints exerted by preceding amino acid positions on sequence composition in T3S effectors.