Table 1.
Average number of reads per sample and average read length of RE-RRS reads.
Table 2.
Hit rates by taxonomic level from RE-RRS samples using one of two restriction enzymes.
Fig 1.
Average abundance (SD) of Hungate1000 Collection genera from the reference-based approach.
Fig 2.
GenBank taxonomies of reference-free tags for ApeKI (a) and PstI (b). Tags were compared against the GenBank database using BLAST and taxonomy was assigned using the MEGAN algorithm considering only hits with the top bitscore for that tag. This figure shows the taxonomy of tags at the kingdom level, and within bacteria and eukaryota at the phylum level and within archaea at the class level. Graphs show the proportion of tags assigned to each taxonomic level and do not reflect the relative abundance of each tag.
Fig 3.
First and second principal components of five metagenome profiling approaches colored by cohort or methane yield.
Metagenome Profiling Approaches included 16S rRNA gene sequencing (a and f), and four restriction enzyme reduced representation sequencing approaches: Reference-Based with the ApeKI (b and g) and PstI (c and h) restriction enzymes, and Reference-Free with the ApeKI (d and i) and PstI (e and j) restriction enzymes. a–e are colored by cohort, with lighter shades of the same color referring to the first sample collected from each sheep and the darker shades referring to the second sample collected from each sheep. f–j are colored by methane yield classification with samples from sheep with low methane yield colored in green and samples from sheep with high methane yield colored in pink.
Table 3.
Comparison of metagenome profiling approaches.
Fig 4.
Compression efficiency of RE-RRS data as the percent of reads sampled decreases.
The compression efficiency of sequence data decreases when less than 5% of reads are sampled, with a sequencing depth that corresponds to 20 times the number of samples sequenced per lane. This number was consistent for both restriction enzymes (ApeKI and PstI) used for this study. Standard errors were 0.000 and are therefore not shown.