Alternative Conformation

[jeleclaire]

PDB and Pubchem files often contain alternative location for atom coordinates (or entire conformation). Mechanisms to switch between these alternative conformations are needed e.g., to generate an alternative atoms data frame.
Particularly, interested with regard to the multiple variants present in the Fragment database.

[jeleclaire]

BiochemicalAlgorithms.jl/src/fileformats/pubchem_json.jl

Lines 527 to 544 in a6eeffc

	for j in eachindex(conformers)
	# Note: the atom will be assigned an id in add_atom!
	atom = (number=compound.atoms.aid[i],
	name="",
	element = isnothing(compound.atoms.element)
	? Elements.Unknown
	: Element(Int(compound.atoms.element[i])),
	atomtype = isnothing(compound.atoms.label)
	? ""
	: compound.atoms.label[i].value, # does the label contain the atom type?
	r = T.(conformers[j][i]),
	v = Vector3(T(0.), T(0.), T(0.)),
	F = Vector3(T(0.), T(0.), T(0.)),
	has_velocity = false,
	has_force = false,
	frame_id = j,
	properties = Properties()
	)

Pubchem files can consists of several compounds which in turn can comprise several alternative conformations. Alternative conformations are stored in thee coordinates section of the file. Currently, our implementation stores different conformation in a molecule by writing all atom coordinates in the dataframe - conformations are distinguished by setting the field frame_id of the atom tuple to the index of the conformation.

However, in PDB files the frame_id is set to the model number:

BiochemicalAlgorithms.jl/src/fileformats/PDB.jl

Line 81 in a6eeffc

atoms.frame_id = orig_df.modelnumber

The model number (or ID) is only available, if all atoms of a structure have an alternative location according to: PDB web site:
In some cases selected residues or parts of residues may have alternate locations as determined by the experiment. Each alternate location of a particular atom is differentiated with a unique Alt ID. For example, the residue number Ser 9 in Chain D in PDB entry 1trz has two atoms, each with alternate IDs A and B. When all the atoms of a structure have multiple locations, they are presented as multiple models and assigned unique Model IDs, often seen in NMR structures (e.g., PDB ID 2kpq).

What exactly should frame_id describe?

BiochemicalAlgorithms.jl/src/core/atom.jl

Lines 3 to 15 in a6eeffc

	const Atom{T} = @NamedTuple begin
	number::Int
	name::String
	element::Element
	atomtype::String
	r::Vector3{T}
	v::Vector3{T}
	F::Vector3{T}
	has_velocity::Bool
	has_force::Bool
	frame_id::Int
	properties::Properties
	end

The different frames of a MD simulation? In this case, I would suggest to add a field conformation_id to the tuple Atom: