Metatranscriptomes from diverse microbial communities: assessment of data reduction techniques for rigorous...

Area of Science:

• Microbial Ecology
• Bioinformatics
• Genomics

Background:

• Metatranscriptome data often contains redundant sequences from diverse microbial populations, necessitating data reduction before annotation.
• Current data reduction techniques, such as clustering and de novo assembly, may impact the accuracy of taxonomic and functional annotation.
• The effect of these techniques on metatranscriptome data annotation and downstream interpretation remains unclear.

Purpose of the Study:

• To investigate the impact of common data reduction techniques on metatranscriptome data annotation.
• To compare the accuracy of clustering and de novo assembly methods for metatranscriptome data.
• To develop a simulation approach for assessing these techniques across varying microbial diversity.

Main Methods:

• Simulated metatranscriptome datasets using 454 and Illumina sequencing technologies with controlled taxonomic diversity.
• Assessed two data reduction methods: clustering and de novo assembly.
• Evaluated annotation accuracy based on protein domain content.

Main Results:

• For Illumina data, a two-step approach (assembly followed by clustering of contigs and unassembled sequences) yielded the most accurate protein domain annotation.
• For 454 data, combining annotations from both contigs and unassembled reads provided the most accurate protein domain annotations.
• The chosen data reduction strategy significantly influenced the accuracy of functional annotation.

Conclusions:

• Assembly should be attempted for metatranscriptome data, even in highly diverse environments.
• Unassembled reads should be included in the final annotation process to improve accuracy.
• These recommendations aim to provide a more accurate reflection of microbial transcriptional activity.