Organizing co mmittee

Indicators 
Score 
Frost resistance 
25 
Conservation of individuals 
10 
Stem formation 
5 
Annual stem growth 
5 
Entering the generative phase 
25 
Possibility of cultivation 
10 
Total points 
80 
Thus, the speed of introduction of 
Silene viridiflora
L. showed that it is possible to 
grow the plant as a source of medicines in the conditions of the Tashkent oasis. Taking 
this into account, we set a goal to study the agrotechnology of its cultivation in the 
future. 

---

Poster presentation 
312 
A MACHINE LEARNING STUDY OF THE ANABOLIC ACTIVITY 
OF ECDYSTEROIDS 
 
S.S. Narzullaev, U.Yu. Yusupova, D.A. Usmanov, N.Sh. Ramazonov
 
S.Yu. Yunusov Institute of the Chemistry of Plant Substances Academy of sciences of the 
Republic of Uzbekistan st. Mirzo-Ulugbek, 77, 100170 Tashkent 
 
In the last decade, medicinal plants have become an important part of the world 
pharmaceutical market. The peculiarity of many drugs is that it is difficult to determine 
the level of pharmacological action of their individual components, because often the 
therapeutic effect of phytopreparations depends on the combined effect of various 
biologically active compounds contained in plant raw materials 
[1]
. 
Ecdysteroids are hormones controlling cell proliferation, growth and the 
developmental cycles of insects and other invertebrates. Recent studies suggest that the 
anabolic effect of ecdysterone, a naturally occurring steroid hormone claimed to 
enhance physical performance, is mediated by estrogen receptor (ER) binding [2]. 
The dataset for the present study has been collected from a series of 23 ecdysteroid 
compounds. AA data (anabolic activity) are taken from our previous study. All original 
in vitro
activity values (5mg/kg) have been converted into molar log(AA) response 
variables.
 
A QSAR study has been performed for the set of 23 ecdysteroids to correlate and 
predict anabolic activity. QSAR modeling was performed applying such methods as GA 
for variable selection among generated and calculated descriptors and MLRA to get a 
final model. Molecular mechanics and quantum-chemical calculations have been 
applied for structure optimization and quantum-chemical properties calculations. 
Three mathematical models for prediction of AA values are proposed. The best 
overall performance is achieved by two-descriptor QSAR model, where 
r
2
values for the 
training and test sets are 0.89 and 0.90, respectively. The significant molecular 
descriptors related to the compounds with AA are: information indices - SIC0, WHIM 
descriptor G3p weighted by polarizability. These variables lead to a molecular level 
explanation of the potency of AA, based on structural descriptors. Obtained model 
(two-variables Model 1) can be used to estimate the anabolic activity of new 
compounds that belong to functionalized ecdysteroids type. 
The following equation represents the two-variable model: 
Log [AA] = 0.8082 (±0.1754) SIC0 - 0.5817(±0.1925) G3p + 6.4905 (±0.1417) (1) 
n
=23; 
r
2
= 0.8899; 
s
= 0.110; 
F
= 64.664; RMSE tr= 0.101; 
q
2
= 0.84 (training set); 
n
=4; 
r
2
ext
= 0.8990; 
s 
= 0.128 (test set) 
According to the obtained data, this model can be used to predict the anabolic 
activity of ecdysteroids.