Compareads: comparing huge metagenomic experiments.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Metagenomic analysis traditionally relies on comparing sequences to existing databases.
• This limits the discovery of unknown or unculturable microbial species, estimated at 99% for Bacteria.

Purpose of the Study:

• Introduce Compareads, a novel de novo comparative metagenomic approach.
• Enable the comparison of metagenomic datasets without relying on a priori knowledge.

Main Methods:

• Compareads identifies similar DNA reads between two metagenomic datasets.
• Utilizes a probabilistic data structure based on Bloom filters for efficient indexing of millions of reads.
• Designed to handle extremely large datasets with limited memory and controlled error rates.

Main Results:

• Compareads successfully retrieves biological information from massive datasets.
• Scalable to over 100 million Illumina reads within hours using approximately 4 GB of memory.
• Demonstrates practical usability on standard personal computers.

Conclusions:

• Compareads offers a practical solution for de novo comparison of large metagenomic samples.
• The developed data structure enables efficient analysis of previously inaccessible biological information.
• The software is freely available under the CeCILL license.