Tutorial on model upload to QSPR-THESAURUS web using JRC models

Tutorial on model upload to QSPR-THESAURUS web using JRC models

1. Find a model in JRC

   • Go to http://qsardb.jrc.it/qmrf
   • Search by 'Ester Papa' (disable search by free text)
   • Select model Q2-17-16-140 QRMF available at http://qsardb.jrc.it/qmrf/search_catalogs.jsp?id=140
   • Save XML file for later use
   • Save pdf file with model description
     
   • You can also download the pdf file here

2. Download data

   • Go to '9.3 Supporting information' section in QRMF form
   • Download both training and test sets (two sdf files)
     
   • If something does not work, download Training and Test set files.

3. Verify or add the article

   • Go to http://www.qspr-thesaurus.eu
   • Login by clicking 'BCF tutorial' link
   • Go to 'Article/Books' section and search for Paola Gramatica article 'An Update of the BCF QSAR Model Based on Theoretical Molecular Descriptors', article A5794
   • The article is there. You can start with next step now.

4. Data upload

   • Go to 'Batch data upload'
   • Select upload one of the sdf files that you downloaded before
   • Inspect preview: select and remap required columns
   • CAS to CASRN (Known column)
   • LogBCF experimental to BCF (Property)
   • add also ARTICLEID by selecting the aforementioned article
   • Start data upload and wait until records are uploaded
   • You will see that records are already in the database (duplicates)
   • (optional) Repeat the same procedure with the Test set
   • If something does not work, upload Excel file with re-mapped columns. Download Excel file

5. Create the training and test sets

   • Go to 'Article/Books' section
   • Select article A5794
   • Select all molecules from the article
   • Create Training set using sort by time
   • Do the same for the Test set
   • If something does not work you can use public sets 'BCF tutorial train' and 'BCF tutorial test' (check 'Show public sets' in browser of sets)

6. Model upload

   • Select the training and test sets
   • Select correct units
   • Select pre-processing steps
   • Select descriptors and 3D pre-procesing (according to your list)
   • Create Excel file with model coefficients as indicated on the tutorial page or download it here
   • Use information from QRMF
   • Upload model
   • Confirm descriptors
   • Wait until descriptors will be shown
   • Correct name of the descriptor
     vIMD,Deg to IDDM
     and repeat all previous steps with model upload again
   • Check AD calculation (leverage)
   • Start calculation of the model
   • Explore the model
   • You can also explore this public model here

7. Add QRMF

   • Go to http://cadaster.eu/QMRF/
   • Select the option 'Upload from an XML file' and click 'Enter the selection'
   • Select the previously downloaded XML file and click 'submit
   • Check the chapters to make sure, everything was parsed in a correct way
   • Click to submit and your QMRF will be stored in the database
   • Use ID on the model page to add QMRF description
   • You can also download the XML file here.