.. |QDESCP_scheme| image:: ../images/QDESCP_scheme.png
   :width: 600

=====================
Descriptor Generation
=====================

In this example we are going to generate a collection of xtb-derived chemical 
descriptors as well as a collection of RDKit-derived descriptors. We 
are going to store them in .json format for each molecule. And we are going to 
create a csv file with the boltzmann averaged values of the descriptors that we 
have calculated per each molecule. The following scheme summarizes the contents 
of this example. 

.. centered:: |QDESCP_scheme|

We are starting from a 'test.csv' file containing the SMILES of the molecules whose 
chemical descriptors we are going to calculate:

.. highlight:: none

::

   SMILES,code_name
   CN1[N]C=NN(C)C1=O,mol_1
   CC1(C)N([O])[CH]N([O])C1(C)C,mol_2
   CC1(C)CCC(C)(C)N1[O],mol_3
   CC1(C)C=CC(C)(C)N1[O],mol_4
   CC1(C)CCCC(C)(C)[N+]1[O-],mol_5

.. highlight:: default

In this case we are going to start by generating some conformers of these 
molecules using rdkit (for more details on the conformer generation please 
check the :doc:`Conformer Search <conformer_search>` section).

.. code:: shell 

   python -m aqme --csearch --input test.csv --program rdkit

Next we proceed to generate the descriptors which is fully automated by the 
QDESCP module. 

.. code:: shell

   python -m aqme --qdescp --program xtb --files "CSEARCH/*.sdf"