Message-ID: <1279592734.41.1632427447239.JavaMail.bigchem@cpu>
Subject: Exported From Confluence
MIME-Version: 1.0
Content-Type: multipart/related;
boundary="----=_Part_40_1256811527.1632427447238"
------=_Part_40_1256811527.1632427447238
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Content-Location: file:///C:/exported.html
1. Can=
I download data?
Yes. Any user can download data that
- are uploaded by him/her (including private data);
- are publicly available and freely downloadable (indicated as: Public and freely downloadable record<=
/strong>). An example of such data are melting point from Bergs=
trom et al article.
- are shared by providing a public id of a model developed (use Download descriptors and model statistic link on the mod=
el page)
2. The f=
irst steps to develop models
The suggested way is to use Comprehensive Modelling interface (see Model=
/Create multiple models menu). Important steps:
- Specify units with physical meaning (e.g., log(mol/l) and not mol=
/l or g/l for model development)
- Select only one method with different (all) descriptors
- one of the longest and CPU consuming part is calculation of descriptors=
; some descriptors such as PyDescriptor take about one minute per molecule<=
/li>
- once descriptors are cached the other methods will reuse them from the =
cache without a need to spend again and again CPU time for their calculatio=
n
- LibSVM requires to select descriptor normalization in advanced options =
(otherwise results with this method will be really bad)
- Do not use the bagging unless you do the final model development. Baggi=
ng will always provide better results than cross-validatio=
n, but requires much more computational resources.
- Identify the best method and several best descriptors and continue work=
with them (outliers exclusion)
- For highly imbalanced classification datasets stratified validation can=
significantly improve the accuracy of models
3. How can I predict new data using an OCHEM model?
Use the Apply model menu
select models that you would like to apply, click "NEXT" at th=
e bottom of page and proceed to upload page, where you can specify molecule=
s to be predicted.
4. Why prediction of even a single molecule could very long?
If prediction of a molecule is cached, the result is shown almost instan=
taneously. The processing of a molecule involves complex steps, such as cal=
culation of descriptors, processing of multiple models, each of which can c=
onsist of hundreds of submodels (for example in case of bagging). All these=
steps can take minutes and sometimes tens of minutes. Sometimes OCHEM is o=
verloaded and the task can wait in a queue before starting calculations. Ho=
wever, if predictions are finished and cached, the result will be instantan=
eous (e.g., for the next prediction of the same structure). The best way - =
predict a set of molecules. The calculations for one and few thousands of m=
olecule take approximately the same time.
5. I have a set of molecules. Sh=
ould I predict them (or calculate descriptors for them) one by one?
No. This is a very inefficient way. The result will be much faster (and =
will require much less computations) if you will send them as a batch (e.g.=
, as an uploaded SDF file with all molecules or using basket).
6.=
Can I calculate and export descriptors?
Yes, unless it was forbidden by the provider of a respective descriptor =
package. These descriptor packages (e.g., Dragon, Adriana) are not listed i=
n the browser for calculation of descriptors.
7. Can I develop models using my own descriptors?
Yes. First you need to upload your structures to OCHEM. This will allow =
you to obtain for each molecule its MOLECULEID in OCHEM (by exporting uploa=
ded structures as a basket). Once it is done, you can map your descriptors =
to the MOLECULEID and upload descriptors to the descr=
iptor storage (see further instructions there). The uploaded descriptor=
s will be shown along with other descriptors on the respective descriptor s=
election page. Naturally, only molecules with uploaded values could be used=
for model development.
8. How can I =
upload my own data?
Yes, of course. You can find instructions how to upload data and interfa=
ce to do it at Batch Upload upload page. See also=
tutorial which shows the required steps.
Briefly, your data should contain (at least) structural information as S=
MILES or SDF, molecular name (if available) and the property data . Select =
the name of the property from the list of available propertie=
s. Also, use units as there were originally provided (it will provide y=
ou a possibility to easily identify and tack errors). OCHEM will provide an=
automatic conversion to any other units, which could be required for model=
ling. Importantly, link your data to an article from which they were upload=
ed. This will again, will help you to be not lost in reviewing them later.<=
/p>
9. I have quantitative data. How can I build a classific=
ation model?
First select records, which you would like to convert to qualitative pro=
perties, to a basket (see t=
his link how to work with data). Click "Edit basket" (e.g., g=
o to Database/Basket and click basket edit ). After this use Discretize the numerical values option to select a =
threshold (or several of them) to create records with qualitative values. T=
hese newly created records can be used for classification models.
Yes, there is an experimental support of this option using all models wh=
ich support MTL (Multi Task Learning) method. However, for this type of ana=
lysis some functionality is not fully supported (e.g., applicability domain=
calculation, etc.) and it is recommended only for exploration purposes.
11. My property is missing in the set of OCHEM =
properties. Can I create a new one?
First of all spend few minutes to find this property. Frequently propert=
y may exist but with a different name, e.g. logKow or Octanol/water partiti=
oning coefficient is named as logPow in ochem. Just try to search for it us=
ing synonymous words. If you found it, there is no need to create it again.=
If property does not exist, you can easily create it. Go to browser of Pro=
perties and click "create new property". After this select its ty=
pe: Numeric for properties which are numeric, e.g. have float values or Cla=
ssification for properties like AMES, which could be "active" or =
"inactive" mutagens. For Numeric property correctly select "=
System of units". If your data are concentrations, select "Concen=
tration". For properties like logPow, which are just ratios or logarit=
hm of ratios, select default "Dimensionless". This is very import=
ant since will allow you to upload data, e.g. in mg/kg but develop models i=
n log(mol/l): OCHEM provides an automatic conversion of units. For classifi=
cation property add available classes, e.g. "active" or "ina=
ctive". Options for classification property should not be float/intege=
r values, e.g. 0/1 will not be allowed by OCHEM. Finally provide some brief=
description of the property.
Download Ames training or test sets from http://ochem.=
eu/article/4027 to see an example how to prepare data for Classificatio=
n properties.
------=_Part_40_1256811527.1632427447238
Content-Type: application/octet-stream
Content-Transfer-Encoding: base64
Content-Location: file:///C:/f10efad794d17bd989ee828baebd9c55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==
------=_Part_40_1256811527.1632427447238--