back to Table of Contents

11. Prediction of pKa's - JOBTYPE pK

JOBTYPE pK calculates the pKs of ionizable sidechains in the TEMPLATE structure. Ionizable groups in proteins have pK's that are often shifted from model compound values (Figure 11.1). The magnitude of the shift from model compound pKs is directly coupled to the electrostatic energy of the group, and is a sensitive probe of the local electrostatic environment (Figure 11.2). The accuracy of pK predictions made using the Generalized Born model within EGAD is reasonable, and gives a similar level of accuracy as Poisson-Boltzmann methods (Figure 11.3; Pokala and Handel 2004).

SCMF calculations are done as a function of pH (1.0 -> 13.5 every 0.25 pH units). Unlike other jobs, both protonated and deprotonated rotamers are explicitly considered for ionizable residues.

The fraction charged for each residue at each pH is listed in the output_prefix.titr file:
pH       residue_id(1) . . . residue_id(n)     total_charge_of_molecule     predicted energy

After the titration is completed, the output_prefix.pK file lists the apparent pKs:
seq_position     residuetype      pK       ∆pK (with respect to model compound pK in resparam file)

The apparent pK is defined as the pH at which the protonated and deprotonated rotamers for a given residue have an equal self-consistent mean-field probability (0.50 in the .titr file). The apparent pK is estimated by linear interpolation between the pH steps flanking the Pcharged = Pneutral = 0.5 point.

In some cases, a residue's pKa is impossible to calculate (especially lys and tyr); these are indicated with not_titr.  

For an example, see examples/pK/barnase.pK.input and the resulting output files.

back to Table of Contents