OK, now that recent stem cell report is really in trouble. One of the main authors, Teruhiko Wakayama, is saying that the papers should be withdrawn. Here's NHK:
Wakayama told NHK he is no longer sure the STAP cells were actually created. He was in charge of important experiments to check the pluripotency of the cells.
He said a change in a specific gene is key proof that the cells are created. He said team members were told before they released the papers that the gene had changed.
Last week, RIKEN disclosed detailed procedures for making STAP cells after outside experts failed to replicate the results outlined in the Nature article.
Wakayama pointed out that in the newly released procedures, RIKEN says this change didn't take place.
He said he reviewed test data submitted to the team's internal meetings and found multiple serious problems, such as questionable images.
These are the sorts of things that really should be ironed out before you make a gigantic scientific splash, you'd think. But I can understand how these things happen, too - a big important result, a groundbreaking discovery, and you think that someone else is probably bound to find the same thing within a month. Within a week. So you'd better publish as fast as you can, unless you feel like being a footnote when the history gets written and the prizes get handed out. There are a few details that need to be filled in? That's OK - just i-dotting and t-crossing, that stuff will be OK. The important thing is the get the discovery out to the world.
But that stuff comes back to bite you, big-time. Andrew Wiles was able to fix his proof of Fermat's Last Theorem post-announcement, but (a) that problem was non-obvious (he didn't know it was there at first), and (b) biology ain't math. Cellular systems are flaky, fluky, and dependent on a lot of variables, some of which you might not even be aware of. An amazing result in an area as tricky as stem cell generation needs a lot of shaking down, and it seems that this one has gotten it. Well, it's getting it now.