Targets

This module contains all the functions to compute target values for molecular structures. The implemented targets at the moment are:

• binary_class: Binary class ID
• capri_class: CAPRI class
• dockQ: DockQ
• rmsd_fnat: CAPRI metric IRMSD, LRMSD or FNAT

As you can see in the source each python file contained a __compute_target__ function. This is the function called in deeprank.generate.

Here are detailed the function in charge of target calculations.

Binary class

deeprank.targets.binary_class.__compute_target__(decoy, targrp)

Calculate binary class ID using IRMSD.

Parameters:

• decoy (bytes) – pdb data of the decoy
• targrp (h5 file hadnle) – HDF5 ‘targets’ group

Examples

>>> f = h5py.File('1LFD.hdf5')
>>> decoy = f['1LFD_9w/complex'][()]
>>> targrp = f['1LFD_9w/targets']

CAPRI class

deeprank.targets.capri_class.__compute_target__(decoy, targrp)

Calculate CAPRI class.

CAPRI class name and ID:

• ‘high’: 0
• ‘medium’: 1
• ‘accepetable’: 2
• ‘incorrect’: 3

Parameters:

• decoy (bytes) – pdb data of the decoy
• targrp (hdf5 file handle) – HDF5 ‘targets’ group

DockQ

deeprank.targets.dockQ.__compute_target__(decoy, targrp)

Calculate DOCKQ.

Parameters:

• decoy (bytes) – pdb data of the decoy
• targrp (hdf5 file handle) – HDF5 ‘targets’ group

RMSDs & FNAT

deeprank.targets.rmsd_fnat.__compute_target__(decoy, targrp, tarname, save_file=False)

Calcuate CAPRI metric IRMSD, LRMSD or FNAT.

Parameters:

• decoy (bytes) – pdb data of the decoy
• targrp (hdf5 file handle) – HDF5 ‘targets’ group
• tarnames (str) – it must be IRMSD, LRMSD or FNAT.
• save_file (bool) – save .izone, .lzone or .ref_pairs file or not, defaults to False.

Returns:

value of IRMSD, LRMSD or FNAT.

Return type:

float

Raises:

• ValueError – Wrong target name
• ValueError – native complex not exist
• ValueError – native complex has empty dataset

Examples

>>> f = h5py.File('1LFD.hdf5')
>>> decoy = f['1LFD_9w/complex'][()]
>>> targrp = f['1LFD_9w/targets']