Icon

Regression QSPR Workflow

General Regression QSPR Framework in KNIME

Connor MacDonald - University of Glasgow

Use of this workflow requires a python environment. A suitable python environment can be prepared using the following guide. In addition, your python environment will require the rdkit package to be installed.

Each step of the QSPR workflow can be probed by investigating the components and metanodes (Access using Ctrl + Double Left Click). Every effort has been made to make the workflow as accessible as possible to chemists with minimal knowledge of QSPRs. To that end, many of the tunable parameters are controlled via component interactive views (Access via Left Click + F10 or Right Click + Select "Open View".

Workflow annotations such as these can be found within each component and metanode to help understand each process in the workflow.

Import

Preparation

Modeling

Reporting

Partioning

Validation

Pre-processing

Hyperparameters
Test Partition Applicability
Feature Importance
Metanode
Test Partition
Validation Partition Applicability
Component
Statistics
Descriptor and Class Selection
Counting Loop Start
Fingerprint Filtering
Partitioned Data Output
Variable Condition Loop End
Compile Report
Data Balancing
Pre-processing
File Selection
Variable Loop End
Descriptor Selection
Generic Loop Start
Class Setup
Filtering Known Compounds
Pre-processing
File Selection
Cross Validation

Nodes

Extensions

Links