ACD/Qualitative Solubility GALAS

Overview

This module classifies the compound into one of the four possible classes according to its solubility in buffer at pH = 7.4 (highly insoluble, insoluble, slightly soluble, soluble). This qualitative solubility assessment is based on the cumulative result of several probabilistic models each predicting the probability of the compound’s solubility in buffer at pH = 7.4 exceeding one of the four established thresholds, ranging from 0.01 mg/ml to 10 mg/ml. Each probabilistic prediction is supported by the corresponding Reliability Index value.

Interface

1. Qualitative estimate of the compound solubility in buffer at pH=7.4 (highly insoluble, insoluble, slightly soluble, soluble) based on cumulative result of several probabilistic models
2. Classification basis - results of individual probabilistic models
3. Up to 5 similar structures from the training set with experimental values

Note: Definition of solubility classes is as follows:

• Highly insoluble – Sw < 0.1 mg/ml
• Insoluble – Sw < 1 mg/ml
• Slightly soluble – Sw > 1 mg/ml
• Soluble – Sw > 10 mg/ml

Technical information

Definition of solubility classes:

Extremely Insoluble		S7.4	< 0.01 mg/ml
Highly Insoluble	0.01 mg/ml <	S7.4	< 0.1 mg/ml
Insoluble	0.1 mg/ml <	S7.4	< 1 mg/ml
Slightly Soluble	1 mg/ml <	S7.4	< 10 mg/ml
Soluble		S7.4	> 10 mg/ml

Training set sizes:

Sub-model and threshold	Number of compounds
non-Extremely Insoluble (S7.4 > 0.01 mg/ml)	5,310
non-Highly Insoluble (S7.4 > 0.1 mg/ml)	5,310
non-Insoluble (S7.4 > 1 mg/ml)	5,692
Soluble (S7.4 > 10 mg/ml)	5,561

Internal validation set sizes:

Sub-model and threshold	Number of compounds
non-Extremely Insoluble (S7.4 > 0.01 mg/ml)	2,277
non-Highly Insoluble (S7.4 > 0.1 mg/ml)	2,277
non-Insoluble (S7.4 > 1 mg/ml)	2,441
Soluble (S7.4 > 10 mg/ml)	2,378

Main sources of experimental data:

• Reference books:
  • The Merck Index. An Encyclopedia of Chemicals, Drugs, and Biologicals, O'Neil, M.J., Smith, A., Heckelman, P.E., Budavari, S., Eds. 13th Edition, Merck & Co., Inc., Whitehouse Station, NJ, 2001
  • Therapeutic Drugs, Dolery, C., Ed. 2nd Edition, Churchill Livingstone, New York, NY, 1999
  • Clarke's Isolation and Identification of Drugs, Moffat, A.C., Jackson, J.V., Moss, M.S., Widdop, B., Eds. 2nd Edition, The Pharmaceutical Press, London, 1986
• Various articles from peer-reviewed scientific journals*

* - Articles reporting solubility models by other authors were the predominant type among analyzed literature, meaning that each publication contained larger collections of experimental data (usually in the order of tens or hundreds compounds) compiled from corresponding original experimental articles.

Internal Validation

Each of the sub-models has been internally validated using their separate internal validation set, constituting ca. 30% of the entire dataset available for a particular threshold model.

Table 1. Performance statistics for the various fractions of the internal validation set of the non-Extremely Insoluble (S_7.4 > 0.01 mg/ml) sub-model of the ACD/Qualitative Solubility predictor.

Subset	Coverage of the entire internal validation set (N=2,277)	Observed*	Calculated probability (p)
			>0.5	<0.5
RI > 0.3 N = 2,146	94.2%	True	1,800 (83.9%)	51 (2.4%)
		False	71 (3.3%)	224 (10.4%)
		Accuracy	94.3%
		Sensitivity	97.2%
		Specificity	75.9%
RI > 0.5 N = 1,800	79.1%	True	1,559 (86.6%)	24 (1.3%)
		False	45 (2.5%)	172 (9.6%)
		Accuracy	96.2%
		Sensitivity	98.5%
		Specificity	79.3%
RI > 0.75 N = 1,054	46.3%	True	936 (88.8%)	5 (0.5%)
		False	11 (1.0%)	102 (9.7%)
		Accuracy	98.5%
		Sensitivity	99.5%
		Specificity	90.3%

* - True means that compound's solubility in buffer at pH=7.4 does exceed the indicated threshold, while False indicates that this parameter is lower than the value indicated in the table name.

Table 2. Performance statistics for the various fractions of the internal validation set of the non-Highly Insoluble (S_7.4 > 0.1 mg/ml) sub-model of the ACD/Qualitative Solubility predictor.

Subset	Coverage of the entire internal validation set (N=2,277)	Observed*	Calculated probability (p)
			>0.5	<0.5
RI > 0.3 N = 2,037	89.5%	True	1,473 (72.3%)	60 (2.9%)
		False	90 (4.4%)	414 (20.3%)
		Accuracy	92.6%
		Sensitivity	96.1%
		Specificity	82.1%
RI > 0.5 N = 1,628	71.5%	True	1,236 (75.9%)	29 (1.8%)
		False	46 (2.8%)	317 (19.5%)
		Accuracy	95.4%
		Sensitivity	97.7%
		Specificity	87.3%
RI > 0.75 N = 908	39.9%	True	725 (79.8%)	4 (0.4%)
		False	9 (1.0%)	170 (18.7%)
		Accuracy	98.6%
		Sensitivity	99.5%
		Specificity	95.0%

* - True means that compound's solubility in buffer at pH=7.4 does exceed the indicated threshold, while False indicates that this parameter is lower than the value indicated in the table name.

Table 3. Performance statistics for the various fractions of the internal validation set of the non-Insoluble (S_7.4 > 1 mg/ml) sub-model of the ACD/Qualitative Solubility predictor.

Subset	Coverage of the entire internal validation set (N=2,441)	Observed*	Calculated probability (p)
			>0.5	<0.5
RI > 0.3 N = 2,153	88.2%	True	1,142 (53.0%)	100 (4.6%)
		False	136 (6.3%)	775 (36.0%)
		Accuracy	89.0%
		Sensitivity	91.9%
		Specificity	85.1%
RI > 0.5 N = 1,634	66.9%	True	918 (56.2%)	47 (2.9%)
		False	67 (4.1%)	602 (36.8%)
		Accuracy	93.0%
		Sensitivity	95.1%
		Specificity	90.0%
RI > 0.75 N = 847	34.7%	True	525 (62.0%)	7 (0.8%)
		False	15 (1.8%)	300 (35.4%)
		Accuracy	97.4%
		Sensitivity	98.7%
		Specificity	95.2%

* - True means that compound's solubility in buffer at pH=7.4 does exceed the indicated threshold, while False indicates that this parameter is lower than the value indicated in the table name.

Table 4. Performance statistics for the various fractions of the internal validation set of the Soluble (S_7.4 > 10 mg/ml) sub-model of the ACD/Qualitative Solubility predictor.

Subset	Coverage of the entire internal validation set (N=2,378)	Observed*	Calculated probability (p)
			>0.5	<0.5
RI > 0.3 N = 2,114	88.9%	True	688 (32.5%)	98 (4.6%)
		False	99 (4.7%)	1,229 (58.1%)
		Accuracy	90.7%
		Sensitivity	87.5%
		Specificity	92.5%
RI > 0.5 N = 1,649	69.3%	True	560 (34.0%)	47 (2.9%)
		False	65 (3.9%)	977 (59.2%)
		Accuracy	93.2%
		Sensitivity	92.3%
		Specificity	93.8%
RI > 0.75 N = 869	36.5%	True	351 (40.4%)	9 (1.0%)
		False	14 (1.6%)	495 (57.0%)
		Accuracy	97.4%
		Sensitivity	97.5%
		Specificity	97.2%

* - True means that compound's solubility in buffer at pH=7.4 does exceed the indicated threshold, while False indicates that this parameter is lower than the value indicated in the table name.