Chemistry Reference
In-Depth Information
5.5 is used to predict log
K
; (ii) for
n
-dentate polyamines forming with metal five-
membered chelate rings, log
K
is proportional to the stability constant (log β
n
) of 1:
n
(M:L) complex with NH
3
:log
K
(polyam ine) = 1.152 log β
n
+ (
n
-1)log 55.5; and (iii) for
more complex systems, some empirical corrections related to chelate ring size, bulk
alkyl substituents, groups bearing neutral oxygen donors, and macrocyclic rings are
made.
The “Stability Constant Estimator” by Hay et al.
10
for 1:1 complexes in water applies
empirical correlations 1-3, equations by Hancock
8
and by Brown-Sylva-Ellis
59
as
well as correlations with some physical properties of metals. This program includes
88 ligands and 74 metal ions. The correlations were trained on 1,431 stability
constants.
The COmplexation of METals (COMET) program
13
applies MLR, ASNN, and
SVM models developed by the ISIDA package.
20
Formally, these models could be
applied for any query ligand. However, a given model is not applied if the query
ligand is too different from the compounds used for the development of the models
and, therefore, is considered to be outside of the model's AD. The program outputs
results of predictions obtained with both the individual and consensus models. The
current version of COMET performs calculations for stability of the 1:1 complexes
of alkaline-earth and lanthanides cations in water at 298 K and ionic strength 0.1
M. The models were trained on 3,340 complexation reactions involving more than
600 ligands. The models for the 1:1 and 1:2 M:L complexes for other metals in water
and in some nonaqueous solvents are in preparation. Potentially, QSPR models link-
ing thermodynamics and kinetic characteristics for solvent extraction could also be
included in COMET.
5.5
IN SILICO
DesIGn oF neW eXtRACtAnts
Theoretical (“
in silico
”) design of compounds possessing the desired properties is
an ultimate objective of modeling. In chemoinformatics, it involves several stages,
shown in Figure 5.12. The sets of carefully selected experimental data (1) are used
1
Database
2
6
QSPR modeling
Synthesis and
experimental tests
in silico
design
3
5
Knowledge base
Virtual screening
4
Generation of
Combinatorial libraries
FIGURe 5.12
Workflow for
in silico
design of new metal binders.
