Fig. 3

A bioinformatics workflow to survey DGR activity and ecology in metagenomes. A Genome map of Trichodesmium erythraeum strain IMS101, indicating the positions of two DGRs, one retron, and several DGR target genes (purple triangles). B An anvi’o [68] visualization of the coverage of a section of the T. erythraeum genome with short reads recruited from the Tara Oceans Project metagenome N000000184, and the distribution of single-nucleotide variants (SNVs). C A detailed representation of the decrease in coverage and increase in the SNV density in Target Gene 1, which indicates the presence of DGR activity in environmental populations. D Distribution of single-amino acid variants (SAAVs) in the same region of the gene. SAAVs are calculated based on the frequency of codons found in metagenomic short reads that fully map to a given codon position in the gene [61], thus, they are much more effective to quantify the extent of non-synonymous variation environmental populations of Trichodesmium have accumulated. E Oligotyping analysis of all metagenomic short reads that fully match to the primer sequence, location of which is indicated by the horizontal bar in panels C and D. The sequence pattern takes into consideration conserved and variable nucleotide positions as revealed by SNVs in read recruitment results. Metagenomic short reads that match the pattern from each sample were subjected to an oligotyping analysis [69], during which high-entropy nucleotide positions divide sequences into distinct, ecologically relevant groups. While the bar plots visualize the oligotype profiles per sample and their diversity, the dendrogram on the left shows how samples relate to each other based on Jaccard Similarity. Finally, arrows point to the sampling sites near Hawaii and Madagascar that correspond to metagenomes that were used in this analysis on a world map