conkit.plot.modelvalidation module

A module to produce a model validation plot

It uses one external program:

map_align for contact map alignment

* This program needs to be installed separately from https://github.com/sokrypton/map_align*

class ModelValidationFigure(model, prediction, sequence, dssp, map_align_exe=None, dist_bins=None, l_factor=0.5, **kwargs)

Bases: conkit.plot.figure.Figure

A Figure object specifc for a model validation. This figure represents the proabbility that each given residue in the model is involved in a model error. This is donw by feeding a trained classfier the differences observed between the predicted distogram and the observed inter-residue contacts and distances at the PDB model.

model

The PDB model that will be validated

Type:Distogram

prediction

The distogram with the residue distance predictions

Type:Distogram

sequence

The sequence of the structure

Type:Sequence

dssp

The DSSP output for the PDB model that will be validated

Type:Bio.PDB.DSSP.DSSP

map_align_exe

The path to map_align executable [default: None]

Type:str

dist_bins

A list of tuples with the boundaries of the distance bins to use in the calculation [default: CASP2 bins]

Type:list, tuple

l_factor

The L/N factor used to filter the contacts before finding the False Negatives [default: 0.5]

Type:float

absent_residues

The residues not observed in the model that will be validated (only if in PDB format)

Type:set

Examples

>>> from Bio.PDB import PDBParser
>>> from Bio.PDB.DSSP import DSSP
>>> p = PDBParser()
>>> structure = p.get_structure('TOXD', 'toxd/toxd.pdb')[0]
>>> dssp = DSSP(structure, 'toxd/toxd.pdb', dssp='mkdssp', acc_array='Wilke')
>>> import conkit
>>> sequence = conkit.io.read('toxd/toxd.fasta', 'fasta').top
>>> model = conkit.io.read('toxd/toxd.pdb', 'pdb').top_map
>>> prediction = conkit.io.read('toxd/toxd.npz', 'rosettanpz').top_map
>>> conkit.plot.ModelValidationFigure(model, prediction, sequence, dssp)

dist_bins

draw()

model

prediction

sequence