openfold.model.heads

Classes

AuxiliaryHeads(config)
DistogramHead(c_z, no_bins, **kwargs)	Computes a distogram probability distribution.
ExperimentallyResolvedHead(c_s, c_out, **kwargs)	For use in computation of "experimentally resolved" loss, subsection 1.9.10
MaskedMSAHead(c_m, c_out, **kwargs)	For use in computation of masked MSA loss, subsection 1.9.9
PerResidueLDDTCaPredictor(no_bins, c_in, ...)
TMScoreHead(c_z, no_bins, **kwargs)	For use in computation of TM-score, subsection 1.9.7

class AuxiliaryHeads(config)

Bases: Module

forward(outputs)

class DistogramHead(c_z, no_bins, **kwargs)

Bases: Module

Computes a distogram probability distribution.

For use in computation of distogram loss, subsection 1.9.8

forward(z)

class ExperimentallyResolvedHead(c_s, c_out, **kwargs)

Bases: Module

For use in computation of “experimentally resolved” loss, subsection 1.9.10

forward(s)

Parameters:

s – [*, N_res, C_s] single embedding

Returns:

[*, N, C_out] logits

class MaskedMSAHead(c_m, c_out, **kwargs)

Bases: Module

For use in computation of masked MSA loss, subsection 1.9.9

forward(m)

Parameters:

m – [*, N_seq, N_res, C_m] MSA embedding

Returns:

[*, N_seq, N_res, C_out] reconstruction

class PerResidueLDDTCaPredictor(no_bins, c_in, c_hidden)

Bases: Module

forward(s)

class TMScoreHead(c_z, no_bins, **kwargs)

Bases: Module

For use in computation of TM-score, subsection 1.9.7

forward(z)

Parameters:

z – [*, N_res, N_res, C_z] pairwise embedding

Returns:

[*, N_res, N_res, no_bins] prediction