Andrew, Did you try the NCS maps in Coot and it didn't work and therefore you are going down this rather convoluted route?
On Fri, Apr 27, 2007 at 10:20:12AM +0100, Eleanor Dodson wrote: > Well - I would do it the same way as you are - find a matrix with > lsqkab, then apply that to the map with maprot. > > It should be OK. > However there are some tricks which might help. Small differences in > cell dimensions sometimes muddle coot I think and also I think it gets > confused by different symmetry in different maps. Hurmpf! I suspect an error with the cells during the maproting. ... but maybe I am biased. One more thing, you can do the lsq match in Coot and get a matrix. You can then apply this matrix to the map in Coot using the transform-map function (that's how I would do it). Paul. > > 1. This shouldnt matter , but I try to make sure I am using coordinates > as close to the origin as I can get them - choose the symmetry operator > to do this. > > If you run pdbset xyzin now.pdb > end > It gives you the centre of mass of the molecule. > > Choose a symmetry operator to apply to move it towards the origin. > > Then get the lsqkab matrix for that pair of solutions. > > 2. I usually expand the maps to P1, make sure I am reading in a piece of > map which completely covers the molecule, and only use P1 symmetry in > coot for that map ( in fact you wont need any symmetry generation ) > > Eleanor > Andrew Robinson wrote: > >Hi all, > > > >I could really use some help with the following situation: > > > >I have electron density maps of my protein in two crystal forms. The > >first is the apo form which is in P43212 with a = b = 88.2, c = 159.1. > >If I soak a substrate for this protein into these crystals, I get a > >second crystal form in P41212 with very similar unit cell dimensions (a > >= b = 88.7, c = 158.7). The two forms are virtually identical except > >that an apparent conformational change causes reverses the handedness of > >the packing. > > > >I'd like to superimpose the maps to in an attempt to find density for a > >ligand, but I'm having a hard time trying to get this to work. The > >closest I have come is by superimposing the coordinates for the two > >forms with lsqkab then using the resulting transformation matrix to skew > >one of the maps. When I check out the skewed map in coot, it is not > >quite right (off by around 2 angstrom). > > > >Is there another way to do this? I'm guessing some of the tricks used in > >multi-crystal averaging would be useful here. > > > >Thanks in advance for your reply, > > > >Andrew Robinson > >Macquarie University, Australia > > > > > >
