Exploiting large-scale drug-protein interaction information for computational drug repurposing

BMC Bioinformatics

Table 6 Comparison of DPIR and chemical fingerprint (similarity) search-based drug repurposing approaches ^a

		DPIR		Similarity
Evaluation of top-ranking scores (%)	Drugs in training set	Fraction	σ	Fraction	σ
High Blood Pressure model
1	1	0.14	0.06	0.09	0.07
5	1	0.36	0.11	0.16	0.07
10	1	0.50	0.14	0.24	0.10
1	2	0.14	0.06	0.14	0.08
5	2	0.48	0.09	0.22	0.09
10	2	0.64	0.09	0.29	0.10
1	3	0.19	0.06	0.18	0.08
5	3	0.53	0.08	0.27	0.10
10	3	0.70	0.08	0.33	0.10
HIV model
1	1	0.16	0.09	0.08	0.07
5	1	0.28	0.14	0.20	0.09
10	1	0.37	0.17	0.25	0.10
1	2	0.17	0.07	0.11	0.07
5	2	0.32	0.12	0.28	0.12
10	2	0.44	0.13	0.36	0.12
1	3	0.19	0.07	0.13	0.07
5	3	0.34	0.11	0.32	0.13
10	3	0.48	0.10	0.44	0.14
Malaria model
1	1	0.14	0.07	0.26	0.18
5	1	0.34	0.14	0.50	0.21
10	1	0.70	0.22	0.55	0.23
1	2	0.20	0.13	0.35	0.16
5	2	0.43	0.20	0.62	0.16
10	2	0.73	0.15	0.72	0.16
1	3	0.21	0.09	0.40	0.14
5	3	0.49	0.15	0.68	0.14
10	3	0.77	0.13	0.79	0.12

^aThe DPIR (type II) models were developed with one, two, or three positive drugs in the positive class of the training set. The chemical fingerprint search used Tanimoto similarity (TS) calculated with the Accelrys ECFP_4 fingerprint. The highest TS coefficient between a baseline compound and a positive compound was used to rank order the baseline compound. DPIR, drug-protein interaction-based repurposing; fraction, fraction of known drugs retrieved; σ, standard deviation; bold font, highest fraction retrieved.

ISSN: 1471-2105