Добавил:
Sekretar
kiopkiopkiop18@yandex.ru
t.me/Prokururor I Вовсе не секретарь, но почту проверяю
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз:
Предмет:
Файл:Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5443_Библиотеки_им_академика_М_И_Перельмана
.pdf
3.7.1.1 Input file and output file detail
The .pdb, .mol, .mol2, .xyz, .sdf, and .smi formats of the chemical are accepted as an
input. Users can also draw the query molecule in the MarvinSketch. As an output, a
quantitative prediction is available in the form of values for mass, hydrogen bond
donor, hydrogen bond acceptors, LOGP, and molar refractivity.
3.7.1.2 How to use?
To screen the molecules for compliance with the Ro5 via the aforementioned tool, one
has to browse an input file in the compatible format at the pH of interest and finally
submit the job to obtain the results as illustrated in Figure 3.3.
Figure 3.3: Steps to screen for molecule compliance with rule of five via SCFBio.
3 The utilization of descriptors in convoluted Lipinski’s rule of five 53
https://t.me/med1917
3.7.2 DruLiTo
DruLiTo is an open-source tool for virtual screening. To predict the drug-likeness, it
filters the molecules for Lipinski’s rule, Veber rule, MDDR-like rule, BBB rule, Ghose
filter, CMC-50-like rule, and quantitative estimate of drug-likeness (QED). DruLiTo
uses the Chemistry Development Kit (CDK), a Java library for descriptor calculation.
This tool is developed by the National Institute of Pharmaceutical Education and Re-
search, NIPER S.A.S. Nagar, Punjab, India. To download this tool, visit http://14.139.57.
41/DruLiToWeb/ DruLiTo_Download.html [57].
3.7.2.1 Input file and output file detail
The.sdfand.molformatsofthechemicalare accepted as an input. Various drug-
likeness filters available in DruLiTo can be selected. As an output, the molecules pass-
ing drug-likeness filter will appear in green color, while the molecules violating drug-
likeness filter will be displayed in pink color.
3.7.2.2 How to use?
To screen the molecules if they comply with the rule of drug-likeness via the afore-
mentioned tool, one has to browse an input file in a compatible format. Later, select
the calculate properties tab to compute all the descriptors. Next, select the drug-
likeness rule box according to the user’s need. The user can even customize the rules
as per the need by clicking on the “customize” button. Finally, click on the “apply fil-
ter” button to apply the selected rules. After the run, “Summary” tab of drug-likeness
rules will appear and one can see the total number of candidate among the molecules
passing and violating all the rules. The molecules that passed the rule can be exported
as .sdf file in the directory of choose. In Figure 3.4, various options to run the job for
predicting the drug-likeness properties of a molecule and an output result file have
been displayed, wherein, (A) supply molecules in .mol or .sdf file format; (B) various
drug-likeness filters available in DruLiTo; (C) example window for customization of
the drug-likeness filter; (D) the molecules passing drug-likeness filter a re shown in
green color, while the molecules violating drug-likeness filter are shown in pink
color; (E) result windows for drug-likeness filter; and (F) the coordinates of molecules
passing the drug-likeness filters can be exported in .sdf file format.
54 Mohit Motiwale et al.
https://t.me/med1917
3.7.3 SwissADME
The SwissADME web tool is a freely accessible tool available at http://www.swissadme.ch
which is well-known for user-friendly submission and easy analysis of the results. It can be
also be used by a nonexpert in CADD. This SwissADME section gives access to five different
rule-based filters, including Ghose (Amgen), Veber (GSK), Egan (Pharmacia), and Muegge
(Bayer) methods. Among these, the Lipinski (Pfizer) filter is the pioneer Ro5. This tool has
been developed by the Swiss Institute of Bioinformatics, Lausanne, Switzerland [58].
3.7.3.1 Input file and output file detail
The 2D structure of the molecule can be drawn in MarvinSketch available as linked tool
with SwissADME, which is later converted to .smi as an input. The output is observed in
descriptors computed for physicochemical properties, lipophilicity, water solubility,
pharmacokinetics, drug-likeness, and medicinal chemistry as displayed in Figure 3.5.
Figure 3.4: Various options to run the job for predicting the drug-likeness properties of a molecule along
with an output result file.
3 The utilization of descriptors in convoluted Lipinski’s rule of five 55
https://t.me/med1917
3.7.3.2 How to use?
The actual input is a list of SMILES as discussed in the above section. Molecules can
either directly be pasted or can be typed in SMILES format, or can be inserted through
the molecular sketcher. The latter allows importing from databases, opening local
files, or drawing 2D chemical structures, converting them to smiles by pressing the
double arrow button. When the list of molecules is ready to be submitted, the user
can start the calculations by clicking on the “run” button.
3.7.4 ADMETLab
ADMETLab is a freely available web server at https://admetmesh.scbdd.com/, devel-
oped by Central South University, China. This web-based platform is developed for
systematic ADMET evaluation of chemicals based on a comprehensively collected
ADMET database consisting of 288,967 entries. It generally predicts 17 physicochemical
properties, 13 medicinal chemistry properties, 23 ADME properties, 27 toxicity end-
points, and 8 toxicophore rules (751 substructures). The physicochemical parameters
Figure 3.5: Results obtained for SwissADME with compliance for drug-likeness rule.
56 Mohit Motiwale et al.
https://t.me/med1917
are enough to distinguish the molecule possessing drug-like properties from those
which do not possess them, based on Lipinski’s Ro5. Not only Lipinski’s but also four
other drug-likeness rules (Ghose, Oprea, Veber, etc.) and one effective drug-likeness
probability prediction model are also included in the drug-likeness evaluation mod-
ule. With a classification accuracy of 0.800 and areas under the receiver operating
characteristic curve (AUROC) of 0.867 by external test set, this model was created
based on MACCS keys using the RF technique. The updated version of it could provide
a more user-friendly design and efficient service, reflected in diverse i nput ap-
proaches, practical explanations, and high computation speed [53].
3.7.4.1 Input file and output file details
This tool accepts the .smi file format for running the prediction for drug-likeness or for
molecule compliance with Lipinski’s Ro5. Alternatively, the molecule can be sketched in
JMSE editor, which is a plugin linked to this tool. However, the input molecule must not
exceed 128 atoms. The output file format is displayed in the form of physicochemical pa-
rameters,i.e.,molecularweight(MW),volume,density,nHA,nHD,nRot,nRing,MaxRing,
nHet, fChar, nRig, flexibility, stereo centers, TPSA, log S,logP,andlogD.Basedoncut-off
for Lipinski’sRo5,thecutofffornHA,nHD,MW,andlogP, drug-likeness can be analyzed.
3.7.4.2 How to use?
Initially, the molecule of interest in SMILES format is submitted at https://admetmesh.
scbdd.com/service/evaluation/index (Figure 3.6). Following submission, the results
will appear in terms of .csv and .pdf formats, which can be downloaded for future
reference. The results obtained after the search have been displayed in Figure 3.7.
Figure 3.6: Input file search box for .smi format available for submission.
3 The utilization of descriptors in convoluted Lipinski’s rule of five 57
https://t.me/med1917
3.7.5 admetSAR
In admetSAR, five classic physicochemical properties, namely the log P, number of hy-
drogen bond acceptors and donors, MW, and TPSA are computed. This tool has been
developed by the East China University of Science and Technology, Shanghai, China.
In admetSAR, over 210,000 ADMET annotated data points have been carefully curated
for more than 96,000 unique compounds with 45 kinds of ADMET-associated proper-
ties, proteins, species, or organisms from a large nu mber of diverse pieces of litera-
ture [59].
3.7.5.1 Input file and output file details
AdmetSAR utilizes .smi file as input to predict the physicochemical properties associ-
ated with the molecule (http://lmmd.ecust.edu.cn/admetsar2). The output file is in
terms of quantitative values for log P, number of hydrogen bond acceptors and do-
nors, MW, and TopoPSA.
Figure 3.7: Result output by ADMETLab.
58 Mohit Motiwale et al.
https://t.me/med1917
3.7.5.2 How to use?
Initially, information associated with the 2D structure of a molecule is fed into admet-
SAR in .smi format. Later, the predict option is selected to compute physicochemical
properties. The results obtained corresponding to each search molecule are displayed
on the screen as in Figure 3.8.
3.7.6 QikProp
QikProp is a one of the modules in Schrödinger, which was designed by Professor Wil-
liam L. Jorgensen. It is brief, accurate, easy-to-use ADME prediction program that
computes physically significant descriptors and pharmaceutically relevant properties
associated with organic molecules, either individually or in batches. Additionally, it
provides ranges for comparing a particular molecule’s properties with those of 95% of
known drugs [60].
3.7.6.1 Input file and output files
QikProp input must be a file containing the 3D structure (x, y, and z coordinates and
atomic numbers) of one or more molecules. Five file formats are accepted:
– Maestro files, uncompressed or compressed (.mae, .maegz, .mae.gz)
– MDL mol files, also known as 3D SD files, uncompressed or compressed (.mol, .
molgz, .sd, .sd.gz, .sdgz, .sdf, .sdf.gz, .sdfgz)
– MOL2 files • PDB files (.pdb)
– BOSS/MCPRO Z-matrix files (.z)
For every completed QikProp job, the files are created in the form of .csv, .qpsa, .out,
and .log.
Figure 3.8: Search tab in admetSAR and prediction space for results.
3 The utilization of descriptors in convoluted Lipinski’s rule of five 59
https://t.me/med1917
3.7.6.2 How to use?
The structures are imported as individual entries with the first molecule displayed in
the workspace. Later, each of the molecules was prepared with LigPrep module of
Schrödinger. Later, QikProp module under application was selected to import the pre-
pared molecules into this panel and finally submit the job to obtain results in the
form of .csv, .qpsa, .out, and .log. The results obtained corresponding to each search
molecule are displayed on the screen as in Figure 3.9.
3.7.7 ADMET descriptors, discovery studio (DS)
DS is comm ercial software, which quantitatively predict properties by a set of rules
that specify ADMET characteristics [61]. It is developed and distributed by Dassault
Systemes BIOVIA (formerly Accelrys). This software can filter hits by applying Lipin-
ski’s Ro5 and Veber’s rules.
3.7.7.1 Input file and output files
DS accepts .sdf,. mol2, and .dsv file formats for an input molecule. The molecule can
be sketched in “sketch and edit molecule” option available with DS. The output is ob-
served in .csv format or .log format.
3.7.7.2 How to use?
Initially the .sdf/.mol2/.dsv file formats of molecules are imported into the DS work-
space. Later, the sketched molecule can be subjected to geometry clean and ligand
preparation using Chemistry at Harvard macromolecular mechanics (CHARMM) force
Figure 3.9: QikProp panel and the physicochemical properties computed by it.
60 Mohit Motiwale et al.
https://t.me/med1917
field. To filter the molecules for Lipinski’s and Veber’s rules, the options displayed in
Figure 3.10 have been set. Finally the job is run.
3.7.8 DrugMint
DrugMint is a web server that assesses a compound’s potential as a medication.
MACCS keys are used to create the prediction model, which is based on the SVM tech-
nique. It forecasts drug-likeness SVM scores for several compounds at once, enabling
users to choose drug-like molecules with ease. The service calculates the compounds’
drug-like SVM ratings from ChEMBL and ZINC. Users may therefore search for poten-
tial molecular drugs by entering their ChEMBL or ZINC IDs. The analog-design module
also offers drug-likeness scores for each molecule and may create a virtual chemical
library using a user-specified scaffold and linkers. This online webserver has been de-
veloped by CSIR, IIMtech India and is available at http://crdd.osdd .net/oscadd/drug-
mint/. Molecular QED scores, which can also be connected, are p redicted using free
QED software. In contrast to the DrugMint model, which is trained for all drug kinds,
QED is used exclusively for oral drugs. DrugMint is a useful free server for choosing
various kinds of drug-like compounds and is not just restricted to oral drug-like mole-
cules due to its simplicity of use and capability to categorize drug-like molecules (ex-
ternal validation accuracy = 0.881) [62].
3.7.8.1 Input file and output file details
.sdf, .mol files are accepted as input files (Figure 3.11A). In the case of absence of such
files, one can draw the structure using Marvin applet. The output is displayed on an
interface as in Figure 3.11B.
Figure 3.10: Options set for filtering molecules based on Veber’s and Lipinski’s rules.
3 The utilization of descriptors in convoluted Lipinski’s rule of five 61
https://t.me/med1917
3.7.8.2 How to use?
This tool allows users to draw the chemical structure using the Marvin applet. Users
have the choice to ei ther build a new molecule or edit/modify an existing molecule.
This module provides the facility to inform the user by email notification after finish-
ing their job. The outcome of the model will be displayed in intuitive interface.
3.7.9 FAF-Drugs4
A new version of FAF-Drugs, FAF-Drugs4 (https://fafdrugs4.rpbs.univ-paris-diderot.fr/),
selects molecules from a huge chemical library that has desirable ADMET characteris-
tics. It conducts computational drug-likeness prediction using PhysChem score, physico-
chemical parameters, and a variety of rule-based systems, such as L ipinski, Veber,
Egan, GSK’s4/400,andPfizer’s 3/75 criteria. Before doing chemical synthesis or testing,
users may choose hits and optimize their strategies by using the summary part of the
filtered findings, which includes information on approved compounds, rejected mole-
cules, and potential causes of outliers. Additionally, it can recognize pan-assay interfer-
ence substances (PAINS) and offer structure alerts. The FAF-QED engine has also been
integrated to generate QED scores (R2weighted = 0.97, R2unweighted = 0.98) and associ-
Figure 3.11: Predicting druglikeness using DrugMint.
62 Mohit Motiwale et al.
https://t.me/med1917
Соседние файлы в папке Библиотека им академика М.И. Перельмана