Similarity Search requires identical column names

Fingerprints as input columns to KNIME’s 'Similarity Search' node are not recognized unless both input columns are named exactly the same.  No errors are given; the column simply doesn't show up in the selectable list if doesn't have the appropriate name (even if the fingerprint types are identical).  This is a minor inconvenience, and a workaround is to rename the fingerprint column before pulling data into this node.  It would be nice if it allowed you to manually select both columns for comparison, the way it is done in chemical similarity nodes from other vendors.



Hi George,

You are right. However, this node also works with other distance measures, which require multiple columns (e.g. Euclidean with 10++ columns). In that case mapping column names is a hard job in the configuration dialog. Since that can also be done upstream of the node we decided to keep the dialog simple. It's already one of the more complex dialogs anyway.