Comparison of bioinformatics workflows (Item P)

Comparison of the performance of the Kraken+Bracken workflow described in Kirstahler et al. 2018, with 12 metagenomic classifiers using a synthetic mock community, as described in Peabody et al., 2015.
Examination of the MetaSimHC250 dataset containing simulated reads from 11 microorganisms with a read length of 250 bp (Peabody et al., 2015). The values obtained using our Kraken+Bracken workflow (see Fig. 2 in Kirstahler et al., 2018) are compared with the values obtained by Peabody et al., 2015.
A) The proportion of assigned reads after classification. Kraken+Bracken performance is similar to Kraken.
B) Calculated sensitivity for tested methods on each individual species in the synthetic mock community. Our workflow shows comparable performance to the other Kraken results.
C) Calculated precision for tested methods on each individual species in the synthetic mock community. Kraken-based methods show high similarities.
D) Bracken results for MetaSimHC250 dataset with calculated sensitivity and precision for the classification of the Kraken+Bracken workflow.

Reference

Kirstahler P,
Bjerrum SS, Friis-Møller A, la Cour M, Aarestrup FM, Westh H., and Pamp SJ. (2018) Genomics-Based Identification of Microorganisms in Human Ocular Body Fluid.

Scientific
Reports, doi:10.1038/s41598-018-22416-4.