Skip to content

Command pca_plot

This command creates a PCA plot of a new data in the NP³ reference chemical space, composed by UNPD+DrugBank+Allosteric datasets (detailed in the Final Reports).

The procedure to create the PCA plots is to first compute the CDK descriptors of a provided list of SMILES, then to create the NP³ reference PCA and to transform these new data to the created chemical space. The resulting PCA is exported to a PNG image.

The provided new data table may contain a column defining the types/categories of each SMILES entry to be used to label the points in the PCA plot.

For processing very big SMILES, the user may manually increase the underline java maximum memory limit size to 4Gb or more (default to 2Gb). In the case of an error "Java Exception ", before executing this command the user may run the following code in the command line: export _JAVA_OPTIONS="-Xmx4g".

Parameters

For the complete list of parameters run:

node np3_workflow.js pca_plot --help

Parameters for the command pca_plot:

  • -t, --table_path <path> : The path to a table in CSV format containing SMILES string in a column. Optionally, it may also contain the types/categories of each SMILES entry in another column. It must be comma "," separated.
  • -s, --smiles_column <name> : The name of the column in table_path containing the SMILES string.
  • -d, --data_type [name] : A name defining the type of the data to be used to label the points or the name of a column from table_path containing the types/categories names of each SMILES entry to be used to label the points in the final PCA plot. This data_type must be different from "UNPD" or "GNPS" and, in the case of a column name, it must contain at most 8 different classes, otherwise only the provided data_type name will be used to label the points. These column values will go to the legend of the PCA plot. (default: "new_data")
  • -o, --output_path [path] : The path to the output directory where the final PCA plots will be stored, together with the descriptors table. If empty string "" (default), the output_path is set to the table_path base directory. (default: "")
  • -n, --output_name [file] : The name to be used to name the final plots, used as a prefix in the PCA plots naming. (default: "new_data")
  • -h, --help : output usage information

Results

In the output_path directory four files are created:

  1. One table with the computed CDK descriptors for each valid SMILES present in the provided table_path, named with the table_path name plus the suffix "_descriptorsCDK.csv";
  2. Two PCA plots containing the reference NP³ dataset used to create the reference chemical space and the new transformed data. These plots are named as: "output_namechemical_space_NP3data_typePCA_scores.png" and "output_name_chemical_space_NP3data_type_PCA_biplot_components.png", this second plot is similar to the first plus the principal components;
  3. The PCA quality of representation circle plot with the reference components named as "pca_quality_representation_cos2_NP3_reference.png".

The third plot result is always the same and depends only on the NP³ reference dataset, used to create the PCA chemical space.

Examples

Example calling the creation of a PCA plot for an external data:

node np3_workflow.js pca_plot --table_path "/path/to/the/table/test_pca.csv" 
--smiles_column SMILES --data_type Libraries 
--output_path "/path/to/the/output/directory/" --output_name test_libraries