So people began to try to work out a stereochemical code and Jeremy Burke began doing site-directed mutagenesis, changing one residue at a time. Changing one residue in the DNA, the DNA target and then looking to see what amino acid he had to do to replace that. Now that seemed to be a very laborious process so at this stage... I decided that wouldn't be the way you could do it, it would be much too complicated, much too difficult because of something that I knew from way back and I mentioned this yesterday, that the structure of DNA was dependent upon the sequence and I realised that the... particularly what brought home to us was that the tightness of packing, so there was a histidine position two of the helix rather than the guanine, those kind of structural chemistry considerations. So that time in the laboratory, and one of the advantages of being here, there's all those sort of things are going on around you and Greg Winter had introduced a way of finding antibodies binding sites by what's called phage display and what's done is a selection method... what's done is if you want to find out... have a particular target and want to find out what the sequences of a structure of a well-known protein which binds to it you would then just make a repertoire, you would change a lot of the binding... you would generate a repertoire of different versions of the same protein and see which one's bound most strongly. We go through several rounds of selection and the way this was done to implement it was done on phage display in other words what you did, this was done by Winter, but based upon the work of a man called Smith who was doing this for peptides rather than for proteins, you had bacterial phage, bacterial virus and you fused to it... you fused to it... the gene or part of the gene for the zinc finger, the DNA sequence for the zinc finger and you made thousands of versions of it. 10⁷ or 10⁸ versions of it and then you had the target and you then selected by infinity, making some conditions more and more stringent and you picked out the selection method. So I decided that, although we talk about making tailor-made proteins to recognise DNA, really, it's a selection process, they're not tailor-made they're selected. And so, the Yen Choo whose name figures here for the first time joined me as a graduate student in about 1990 and I put him on to start off the site-directed mutagenesis and that was clear that this was going to take a very long time. And then Greg Winter was doing this is the lab so I thought that we would really do it by phage display. Now phage display is all right, you put an antibody, so you see you have the sequence... let's go back to phage display with an antibody, you have the antibody at the tip of the bacterial virus and you have the gene inside. You have the DNA sequence for that inside, so it's like a B-cell in immunology. So you select, you select on the phage, and with a large library, and then you find which phage's bind and then you look at the sequence of the DNA inside that tells you what the amino acid sequence is. So that's a very beautiful system. Sydney Brenner was trying, also jumped into this but he was trying to do it with lambda and other people tried to do it with whole cells. This was called repertoire selection technology, it's very standard now, but we were fairly early on this.