TIDEA

TIDEA (Target Independent Drug Enhancement Algorithm) is an in-silico screening computer program that distinguishes potent small molecule inhibitors from weaker inhibitors that bind the same target. In sharp contrast to conventional QSAR and docking-based approaches, TIDEA requires only the 2D structure of the small molecule. TIDEA predicts relative adhesiveness rather than shape complimentarity: it does not require knowledge of the target structure, target identity, or bioactivity of related small molecules (SAR). Selection of small molecules with high TIDEA scores increases average potency and drug-like character without decreasing small molecule structural diversity or target diversity, making it ideal for initial filtering of universal libraries and corporate compound collections prior to biological screening.

A Poster presentation on TIDEA was made by Darryl Rideout at the 241st ACS National Meeting & Exposition in Anaheim California on 3/29/2011. Please click here to open a copy of the entire poster as a .pdf file. A portion of the poster follows:

TIDEA (Target-Independent Drug Enhancement Algorithm) is a proprietary algorithm used to estimate potency and hit rates for small molecule ligand/protein target interactions from ligand structure alone, without any knowledge of target structure or SAR. We used a Learning Set of 120 diverse bioactive ligands with known potency values for 56 distinct protein targets to develop TIDEA. The TIDEA algorithm calculates a score solely from the ligand structure, which correlates with potency (pIC50/pKi) for a wide array of ligand/target interactions. In a Test Set of 80 ligands with an average of 4 ligands/target (i.e. 20 targets), which had no overlap with the Learning Set, the percentage of subnanomolar ligands was 11-fold greater for high TIDEA scores (>9.5). with statistical significance (Chi Square p value <0.01). When applied to an Ultradiverse Set of 65 ligands with a different target for every ligand, statistical significance was observed for the percentage of potent (< 100nM) ligands (Chi Square p value < 0.03), and for the increase in average potency (T-Test p value < 0.015). TIDEA has the potential to accelerate drug discovery by addressing key limitations of structure-based and SAR-based methods, because TIDEA shows a selection bias for potent molecules while maintaining diversity.

Traditional methods of achieving selection bias toward more potent small molecules (docking, QSAR, analog-based approaches, etc (1-7)) are limited in that they require target macromolecule structure or SAR knowledge, and because they restrict molecules diversity. TIDEA (Target Independent Drug Enhancement Algorithm), a virtual screening algorithm for bioactive small molecules that uses a proprietary algorithm developed at Focus Synthesis LLC, requires no knowledge of the target macromolecule or active site is required. Ligand potency (ClogIC50 or ClogKi) increases with increasing TIDEA scores for a wide array of ligands with scaffolds and diverse targets. This TIDEA score/potency correlation is statistically significant. TIDEA correlates with adhesiveness potential for bioactive ligands that fit some macromolecular binding site, rather than specific shape complementarity addressed by traditional approaches. TIDEA has the potential to increase the success rate in early stage drug discovery and development and decreasing late stage failures by providing a broader range of viable drug candidates in early discovery.

Learning Set for Creation of the TIDEA algorithm. (See also Scheme 1). The Learning Set used to develop TIDEA was constructed from 56 distinct subsets of 2 to 4 ligands per subset reported in the literature, for a total of 120 small molecule ligands (8). Each of the 56 ligand subsets has a distinct scaffold and binds to a distinct macromolecular target, to ensure that potency/TIDEA score relationships would not be related to specific ligand/target shape complementarity.

Test Set. The Test Set consists of 80 diverse small molecule ligands (FW between 300 and 500) that bind 20 distinct targets (8). We deliberately avoided overlap between the 56 Learning Set targets and the 20 Test Set targets to ensure a rigorous test of the TIDEA technology. The ligand scaffold types are also distinct for the learning set and test set.

TIDEA effectiveness for the Test Set: There is a significant correlation between TIDEA score and potency for the Test Set, as shown in Figure 1. Seven out of eight macromolecular ligands have TIDEA scores greater than 9.5 in the Test Set. Average Potency increases disproportionately with TIDEA score than with molecular weight, as shown in Figure 2. TIDEA scores above 11.5 yield a significantly larger number of potent (< 5 nM) inhibitors without compromising diversity, as shown in Figure 3. Statistical significance. The difference between the average potency (ClogP or ClogKi) for TIDEA score <9.5 (7.11) and TIDEA score >9.5 (7.89) is statistically significant by the T test (p<0.01). The percentage of subnanomolar inhibitors is increased more than 10 –fold when the higher (TIDEA score >9.5, 22% subnanomolar) score range is compared to the lower (TIDEA score < 9.5, 2% subnanomolar). This enrichment of subnanomolar inhibitors is statistically significant by the Chi Square method (Yates p-value < 0.01), as shown here: