Fig. 4

CORRAL yields high sensitivity and specificity when predicting the presence of eukaryotes in metagenomic data. A Proportion of results (Y-axis) yielding the correct species at a lower limit of detection for EukDetect and CORRAL. B Unrepresented species simulation results. Number of samples (count; Y-axis) is shown for which results were produced by EukDetect or CORRAL (X-axis). C, D ZymoBIOMICS mock community standard from Yang et al. [21]. C Heatmap showing copies per million (CPM) for all microbial eukaryotes detected by CORRAL in 6 replicate samples of the ZymoBIOMICS standard extracted using the ZYMO protocol as described previously [21]. Asterisks mark unambiguous results from CORRAL. D Plots showing the ratio of reads assigned to S. cerevisiae versus C. neoformans for CORRAL, EukDetect, “no filter” and “MAPQ filter” (see the “Methods” section). Separate box plots are shown for each extraction method used, including MagPure Fast Stool DNA KF Kit B (“MP”); Macherey Nagel NucleoSpin Soil kit (“MN”); Zymo Research Quick-DNA Fecal/Soil Microbe kit (“ZYMO”); protocol Q (“Q”); MOBIO DNeasy PowerSoil kit (“PS”); and a non–kit-based manual protocol adopted by the Metagenomics of the Human Intestinal Tract Consortium (“MetaHIT”). Each point represents a separate mock community sample that was extracted, sequenced, and analyzed