I am processing negative stain data of my protein bound to antibody. I can see 2d classes with protein bound to one Ab, protein bound to 2 Abs and protein bound to 3 Abs.
But when I do an abinitio (121163 particles were input to abinitio) with multiple classes, I am only able to get a volume of protein bound to one Ab. I also tried a single class abinitio. But I get the same result.
Set aside the particles from that class (for only this task) and run again. Also do (number of expected conformers + 1) classes so it has somewhere to store the junk particles. Or take all particles and run classification with several copies of the 1 good ab initio model. It should work it out. Or temporarily select only the 2d classes that show what you want and try with those. Careful as these approaches lend themselves to user bias