Mike Jackson’s account of the early molecular work at IRRI’s International Rice Genebank is a nice reminder that the idea of the “genomic genebank” (as he calls it) is not as new as it may sounds. In the 1990s, RAPD and AFLP markers were already being used to identify duplicates, reveal genetic structure and, more ambitiously, predict which accessions might possess useful traits. It’s the continuation of the trajectory I traced in my recent post on descriptors: from names and human-scored traits to photographs, digital phenotypes and now genomic information, each adding a layer of information that makes the collection more searchable and usable. The question keeps shifting from “what do we have?” to “which of what we have might be useful?”
Jackson’s story is also a key piece of the argument I tried to make in another recent post: preserving options is only the beginning. More, and better, information makes options easier to discover, but discovering an option is not the same as exercising it. The harder question is what happens next: how do we turn knowledge about what’s in a collection into actual selection, testing, breeding, adoption and impact? Data can open the door to better use of genebank collections. It cannot walk through it for us.
And there is a danger here. As our information about genebank collections becomes ever more layered, richer and more precise, it can start to look as though we’re solving the problem of use. We are not. We are solving the problem of finding possibilities. That’s only one part of the journey. A genomic prediction is not a breeding line; a photograph is not a phenotype under farmers’ conditions. The distance between knowing an option exists and actually exercising it still has to be travelled. That is a social and institutional process as much as a biological and technological one, and it starts only when the search is over.