The problem. Binning produces draft genomes, but some are missing half their content and some are two organisms merged by mistake. Before you build on a metagenome-assembled genome you need to know two things: how complete it is, and how contaminated. Without a reference for an uncultured organism, there is nothing obvious to compare against, so the quality question looks unanswerable.

The idea. CheckM uses marker genes that are expected in single copy across a lineage. It places a genome on a reference tree, picks the set of single-copy markers appropriate to that lineage, and counts them. Markers that are present measure completeness. Markers that appear more than once measure contamination, since a clean single genome should carry each exactly once. Reporting both as percentages gives every draft a pair of numbers that the field now quotes as standard.

Why it matters. These two numbers are the gate that separates a usable genome from an artifact, and they made large MAG collections filterable and comparable. Using conserved single-copy genes as an internal ruler is a clean idea that reappears wherever you need to judge completeness without a reference.

Verdict. The quality check every MAG passes through. Read it for the lineage-specific single-copy marker sets.