Table 1.

The most common functional group filters described in the scientific literature presented in alphabetical order.

Name/Reference	Description	Features/Cutoff Values
Aggregators [36]	Tanimoto coefficient similarity search to a database of known aggregators.	Tanimoto coefficient similarity ≥ 0.85 or SlogP > 5 (high similarity),
		Tanimoto coefficient similarity ≥ 0.5 and SlogP > 3 (medium similarity),
		Tanimoto coefficient similarity < 0.85 and SlogP ≤ 3 (low similarity)
Ely Lilly Rules [37]	A set of 275 rules, developed over an 18-year period, used to identify compounds that may interfere with biological assays, allowing their removal from screening sets.	Reasons for rejection of compounds: reactivity, interference with assay measurements (fluorescence, absorbance, quenching), instability and lack of druggability (lacking both oxygen and nitrogen)
Muegge method [18,38]	Bioavailability prediction rules dubbed the Muegge method. Pharmacophore filter developed by analyzing known drug databases, with four functional molecular motifs determined to be important in drug-like molecules:	Primary, secondary, and tertiary amines are considered pharmacophore points but not pyrrole, indole, thiazole, isoxazole, other azoles, or diazines. Compounds with more than one carboxylic acid are dismissed. Compounds without a ring structure are dismissed. Intracyclic amines that occur in the same ring are fused and count as only one pharmacophore point.
PAINS [39]	Removal of frequent hitters (promiscuous compounds) by identifying sub-structural features not recognized by filters commonly used to identify reactive compounds.	Functional groups such as rhodanines, phenolic Mannich bases, hydroxyphenylhydrazones, alkylidene barbiturates, alkylidene heterocycles, 1,2,3-aralkylpyrroles, activated benzofurazans, 2-amino-3-carbonylthiophenes, catechols, and quinones do not pass the filters.
REOS 1 [3,40,41]	Seven property filters	H-bond donor ≤ 5,
	(similar to the PATTY	H-bond acceptors ≤ 10,
	rules in program developed at Merck)	−2 ≤ Formal charge ≤ +2,
		Number of rotatable bonds ≤ 8,
		200 ≤ Molecular weight ≤ 500,
		20 ≤ number of heavy atoms ≤ 50,
		−2 ≤ logP ≤ 5
	Functional group filters for the removal of problematic structures dubbed REOS (rapid elimination of swill; program developed at Vertex).	Reactive, toxic and other undesirable moieties such as nitro groups, preoxides, triflates, aldehydes, acetals, etc.

¹ REOS is a hybrid filter which combines a set of functional group filters with property filters. As the REOS filter can be combined with other (property) filtering schemes, the property filtering part can be omitted and only functional group filters employed. As implemented in KNIME, the user can also specify the maximum quantity for each of the functional group rules, tuning the filter to the needs of the individual research scenario. REOS moieties in the SMARTS format can be found inside the KNIME workflow “REOS substructures” node.