Kraken and Braken – Possible Wonder Twins of Metagenome Analysis
Contributed by Marina Becker
For the last several months, the Weimer lab has been on a push to sample, sequence, analyze and interpret a variety of different meta-genomic samples in hopes of gaining insight into the variability of the microbiome between individuals, physiological locations, and treatment conditions. In order to make the most of our sampling resources, we have developed a fairly standard pipeline for these types of analysis. Standardizing workflows for a certain sample type can greatly improve analysis efficiency and can help with debugging both the computational and wet lab sides of an experiment. Knowing what to expect of the analysis of a sample whose content is somewhat unpredictable helps to reduce at least one source of variation. All that being said, it is important to set your standard with a software and a process you trust. For us, that is where Kraken coms in.
Kraken was developed at John’s Hopkins for the purpose of read classification in a meta-genomic sequence pool. It takes a k-mer-based searching approach that looks for unique kmers in a read compared to a database and assigns the read to a certain taxonomic level. The program is fast, reasonably light, and accurate. It also has multiple run modes for optimization in a particular setting. Kraken does not come without its caveats, though. For instance, when running in “fast” or “mini” mode you may sacrifice accuracy or performance. It is also worth noting that the database that acts as a kmer reference ultimately determines the content of your outputs (meaning that if an organism is not in the database, it won’t be recognized in the sample, even if it is actually there). However, the customizability of the program makes it versatile for users with varying types and levels of resources. On the whole, we have found the outputs to be consistent, reliable, and easy to both interpret and feed into other downstream analyses.
In recent news, Johns Hopkins has announced the release of a sister software to Kraken called Braken. This supplement to the original software is for more accurate prediction of species abundance in a meta-genome – a feature the original Kraken was only able to touch on with k-mer hit counts. We have yet to test the new software but are looking forward to exploring its potential. In the meantime, we will continue to use Kraken as our go-to guy for meta-genome read classification.
Read more about Kraken here:
And Braken here: