|
Table 1. Comparison of Calculated Log P Values with Experiment
Compound
(plant sourse) |
LogPexpa
|
ACDLabs/LogP
|
CLogPe
|
CACheLogPf
|
| phenol |
1.42
(1.49)b |
1.48 |
1.47 |
1.38 |
| o-cresol |
2.03
(1.95)b |
1.94 |
1.97 |
1.66 |
| p-chlorophenol |
2.28
(2.39)b |
2.43 |
2.48 |
2.87 |
| bromobenzene |
3.18
(2.99)b |
2.99 |
3.01 |
3.26 |
| naphthalene |
3.46
(3.37)b |
3.45 |
3.32 |
3.45 |
| biphenyl |
3.84
(4.04)b |
3.98 |
4.03 |
4.08 |
| 1
(soybeans) |
2.74
(2.62)c |
3.15 |
3.17 |
3.28 |
| 2 (soybeans) |
2.44d |
2.82 |
2.49 |
3.37 |
| 3
(kidney beans) |
3.59
(3.58)c |
4.22 |
3.96 |
3.64 |
| 4
(potatoes) |
3.13d |
3.27 |
2.63 |
2.76 |
| 5
(potatoes) |
3.26d |
3.13 |
2.46 |
2.59 |
| 6 |
2.88 |
6.51 |
6.24 |
3.78 |
| 7 |
2.90 |
6.50 |
6.33 |
4.40 |
| 8 |
3.37 |
6.93 |
6.65 |
4.27 |
| 9 |
3.34 |
7.47 |
7.18 |
4.61 |
| 10
(broad beans) |
|
3.08 |
2.97 |
3.73 |
| 11 |
|
3.83 |
4.03 |
3.87 |
| 12 |
|
3.07 |
3.72 |
3.50 |
| 13 |
|
3.31 |
3.35 |
2.99 |
| 14 |
|
3.16 |
3.56 |
3.52 |
a Data taken from Spessard, G. O.; Matthews, D. R.; Nelson, M. D.; Rajtora, T.
C.; Fossum, M. J.; Giannini, J. L. J. Agric. Food Chem. 1995, 43, 1690; measured by
reverse phase HPLC. b Data taken from Nys, G.G; Rekker, R. F. Eur. J. Med Chem.
Chim. Ther. 1974, 9, 361. c Data taken from Arnoldi, A; Merlini, L. J.
Agric. Food Chem. 1990, 38, 834; measured by reverse phase HPLC. d Data
taken from Fossum, M. J. Research report to Gary O. Spessard, 1993. e CLogP is
a program produced by BioByte Corp., Claremont, CA. f CAChe LogP is a protocol
for calculating log P values using CAChe ProjectLeader software developed by the Oxford
Molecular Group, Beaverton, OR.
The first six compounds listed in Table l are standards whose log P values have been
carefully measured and range from ca. 1.5 to 4. Within this spread of values are the log
Ps of a large number of naturally occurring molecules. All three programs seem to
calculate log Ps in good agreement with experiment. For compounds 1 - 5, all three
programs also give reasonable agreement with each other and with experiment, the only
exception being the value for compound 3 calculated by ACDLabs/LogP, which seems rather
high.
The abietic acid series (compounds 6 - 9) shows wide disparity between calculation
and experiment. ACDLabs/ LogP and CLogP both give values of log P that are up to three
logarithmic units higher than experimental results, while CACheLogP gives results that are
much closer to experiment. That ACDLab/LogP and CLogP give similar numbers is perhaps not
surprising, since both use similar computational approaches. In contrast, the CACheLogP
protocol is based upon SCF semiempirical quantum mechanical computations and not
incremental contributions of fragments and atoms. CACheLogP values are determined as a
function of the difference in heat of formation of solute in the gas phase and water phase
as well as the calculated solvent-accessible surface area. A training set of over 200
compounds, whose log P values are known from experiment, was used in developing the
protocol.
Compounds 10 - 14 are either naturally occurring (10) or analogues of phytoalexins
isolated from the broad bean plant that have been synthesized by my research group
(11 - 14). Although the log P values of these compounds have not yet been determined
experimentally, it is nice to see that all three software packages give similar results.
ACDLabs/LogP db 3.5 is an excellent program for calculating log P values, in most cases
providing reliable estimates of this most important chemical property. Like all such
programs, however, its effectiveness must always be judged in comparison with experiment.
Figure 1. Phytoalexins and analogs
|
11
12: R1=O, R2=Me