MoSS

This node searches for frequent molecular fragments in a set of molecules. The algorithm used is Christian Borgelt's MoSS implementation.

Options

Column with molecule structure

Select the column in the input table that contains the molecule strings (Smiles, SLN or SDF).

Column with class labels

Select the column in the input table that contains the class labels (must be a string column). Use the twin-lists below to assign each class label value to the active (focus) or inactive (complement) group used during the search.

Focus labels

Select the class label values that form the active (focus) group. The included values will be treated as active molecules during fragment mining.

• Class label: The class label value
• Focus label: Group labels for fragment mining.
  • Ignore: Molecules with these class labels are ignored, i.e. not considered part of the focus or complement.
  • Active: Molecules with these class labels are considered active, i.e. part of the focus.
  • Inactive: Molecules with these class labels are considered inactive, i.e. part of the complement.

Minimum focus support (%)

Sets the minimum number of fragments — a fraction of the number of input molecules in the active class — a fragment must occur in order to be frequent and thus reported.

Maximum complement support (%)

Sets the maximum number of fragments — a fraction of the number of input molecules in the inactive class — a fragment may occur in order to be reported.

Minimum fragment size

The minimum size (number of bonds) a fragment must have in order to be reported.

Maximum fragment size

The maximum size (number of bonds) a fragment may have in order to be reported.

Ignore pure carbon fragments

Prevents the search for fragments that only consist of carbon atoms.

Use ring mining

If enabled, rings of the specified sizes are treated as single entities which makes the search much faster and also avoids finding fragments with partial rings.

Minimum ring size

The lower bound for ring sizes that are treated as single entities during ring mining.

Maximum ring size

The upper bound for ring sizes that are treated as single entities during ring mining.

Start with core

Provide a SMILES string that is used as a seed for starting the search. All found fragments will contain this core structure.

Canonical form pruning

Turns canonical form pruning on or off. Canonical form pruning reduces the search space by only extending canonical representatives of equivalent fragments.

Equivalent sibling pruning

Turns equivalent sibling pruning on or off. Equivalent sibling pruning avoids redundant extension steps for equivalent fragment siblings in the search tree.

Perfect extension pruning

Turns perfect extension pruning on or off. Perfect extension pruning cuts the search when a fragment cannot possibly yield frequent extensions.

Find chains of variable lengths

Enables the search for fragments that are equal except the length of their carbon chains.

Fragment extension mode

Choose between the two possible ways to extend a fragment during the search.

• Maximum source extension: Extends fragments by attaching new atoms/bonds to the maximum source node.
• Right-most path extension: Extends fragments along the right-most path in the DFS tree.

Maximum embeddings used (0 = unlimited)

Sets the upper limit on the number of embeddings that are stored in memory. Set to 0 for no limit.

Views

This node has no views

Workflows

• 01_Example_for_using_MoSSKNIME Hub
• 01_Example_for_using_MoSSKNIME Hub
• 01_Example_for_using_MoSSKNIME Hub
• 01_Example_for_using_MoSSKNIME Hub
• 05_DrugBank_Network_analysisKNIME Hub

Links

• No links available

Developers