A tandem regression-outlier analysis of a ligand cellular system for key structural modifications around ligand binding

Table 1 Dominant descriptors for each data size and the frequency of Jurs_RNCG (Jurs type descriptors) throughout 521 data sets are summarized here

Data size	Dominant descriptor (Near dominant)	Frequency of Jurs descriptors (Jurs_RNCG/Jurs type /521 data sets)
10	Molecular_PolarSASA	4/41/521
20	Molecular_FractionalPolarSASA	5/25/521
30	ES_Count_dO^b	4/14/521
40	Num_RingBonds	10/14/521
50	SC_3_P^a	9/16/521
60	SC_3_P^a	9/12/521
70	IC^a	18/31/521
80	IC^a	18/33/521
90	Jurs_RNCG	21/31/521
100	IC^a (Jurs_RNCG)	19/31/521
110	Jurs_RNCG	16/21/521
120	Num_AtomClasses (Jurs_RNCG)	21/23/521
130	IC^a (Jurs_RNCG)	28/29/521
140	IC^a (Jurs_RNCG)	39/306/521
150	Jurs_RNCG	39/39/521
160	IC^a (Jurs_RNCG)	53/53/521
170	Jurs_RNCG	369/369/521

^a The descriptors SC_3_P and IC are calculated based on graph theory.
^b The descriptor ES_Count_dO is a fractional descriptor (The definition of a fractional descriptor is in the discussion section).
Agonists of each data set are selected out of the 178 carboxylic acid PPARγ agonists. The top-ranked descriptor of each set was selected from 521 descriptors of eminent classes.