Functional principal component data analysis: a new method for analysing microbial community fingerprints.

Area of Science:

• Microbiology
• Bioinformatics
• Statistical Modeling

Background:

• Molecular characterization of microbial communities often uses small subunit (SSU) rRNA gene amplification and fingerprinting techniques.
• Existing statistical models, like principal component analysis (PCA), have limitations in distinguishing closely related bands and analyzing gel structures.
• Current methods often ignore the local neighborhood structures within denaturing gradient gel electrophoresis (DGGE) gels.

Purpose of the Study:

• To assess if functional data analysis (FDA) methods can improve the discriminatory ability of molecular fingerprinting techniques.
• To compare the effectiveness of FDA against standard PCA for analyzing microbial community banding patterns.
• To introduce a novel statistical approach that incorporates neighborhood structures in gel-based analyses.

Main Methods:

• Applied functional data analysis (FDA) methods to molecular fingerprinting data.
• Treated band intensities as a function of their position on the gel, including neighborhood structures.
• Conducted a simulation study to compare FDA with principal component analysis (PCA).
• Analyzed experimental denaturing gradient gel electrophoresis (DGGE) data using the FDA approach.

Main Results:

• A simulation study demonstrated the weaknesses of standard PCA compared to the FDA approach.
• FDA showed improved discriminatory ability in analyzing microbial community banding patterns.
• The FDA approach effectively analyzed experimental DGGE data by considering local gel structures.

Conclusions:

• Functional data analysis (FDA) provides a more powerful and discriminatory method for analyzing microbial community molecular fingerprints.
• FDA overcomes limitations of traditional methods like PCA by incorporating spatial information from gel banding patterns.
• This approach enhances the accuracy of microbial community characterization, especially in complex samples.