Also a neat site. It seems like the ratio of usable protein would be a good addition to, but a lacking substitute for, the raw quantities. Especially given how incomplete the PCDAAS and DIAAS datasets are.