demianriccardi/crysenm

Fortran 90 toolkit for application of elastic network models to proteins in the crystalline state

crysenm's Issues

potential improvement: reduce dependency on CrysFML

CrysFML is great and easy to install, but CrysENM could be called on pdbs with the associated symmetry operators. I would be a matter of parsing the pdb and pulling them out.

Protein_atom_list_type/Atom_list_type and protein_atom_type/atom_type Mess

New types were introduced to have protein atom attributes. The protein_atom_type extends the atom_type from CFML. From Modern Fortran Explained, any subroutine that expects the atom_type can be called with the protein_atom_type by passing the variable%atom_type to the sub. This seems straightforward. However, we need also need a protein_atom_list_type, which is analogous to the atom_list_type from CFML. Need to figure out how to use parent class of atom_list_type from a protein_list_type. Do I need to separate those attributes and just index them properly?

filling the unit cell

My code depends on being able to reduce the symmetry and fill the primitive unit cell. This isn’t so bad for proteins, because of their symmetry limitations, and the following double loop (print out the cartesian xyzs for all atoms in a unit cell) will generally work:

type (space_group_type) :: SpG
type (atom_list_type) :: asym_un
type (Crystal_Cell_Type) :: Cell

!.. After reading in the asymmetric unit in fractional coordinates from a cfl file…

do j=1,spg%multip
do i=1, asym_un%natoms
call Get_Orbit(asym_un%atom(i)%x(:),SpG,nmultip,orb, preserve=.true.)
x(:) = matmul( orb(:,j), transpose(Cell%Cr_orth_cel)) ! Switch the cartesian
print '', asym_un%atom(i)%lab, x(:)
end do
end do

The loop above is not so efficient as it calculates the orb variable containing equivalent positions each time, but it keeps things ordered in the print. I would like to make this work more generally, but I need to be able to collapse down on redundant coordinates generated from Wyckoff positions. Is there a straightforward way of doing this in CFML?

make the CrysENM encapsulate CrysFML

programs using the cysenm should be able to do useful things with minimal lines of code. The goal is to not have to access CrysFML directly. The target of CrysENM is much more focused than that of CrysFML:

ENM of crystalline molecules.
Multi scale models of Diffuse X-ray scatter.