A lab robot counts molecules, and gets it wrong. One tube shows 0, another 12, a third 3, though in reality they all hold roughly the same amount. Such 'raw' numbers are common in biology: DNA sequencing, counting species in a water sample, gene activity. There's plenty of data, but it's sparse and more scattered than simple randomness would predict.

Mathematicians have proposed a 'translator' for this noise — a residual method based on the Dirichlet-multinomial (DM) distribution. The idea is like a spell-checker in a word processor: it doesn't rewrite the text from scratch, just underlines suspicious spots, leaving everything else untouched. Same here: the method doesn't alter the original zeros and ones; it only compresses repeated values if they stand out from the overall pattern. The entire cleanup happens in one pass, and processing time depends only on the number of nonzero cells, so it's nearly instantaneous.

The new approach is like a universal adapter. It works not only with simple tables but also with data where features are linked in chains or trees (e.g., species on a phylogenetic tree). This lets you clean data about genes, proteins, or ecosystems with the same tool. And crucially, you can just 'plug it into' existing analysis programs without rewriting them entirely.

The corrector doesn't invent facts. It just removes the ripple, leaving the text, or the table, cleaner than before.