Lhasa Limited shared knowledge shared progress

Negative Predictions

Derek Nexus contains expert-derived functionality to provide negative predictions for bacterial in vitro mutagenicity and skin sensitisation, instead of previously generated "nothing to report" outcomes.

For further information regarding Negative Predictions and how this has been implemented within Derek Nexus, please take a look at a paper written by Richard Williams et al. in Regulatory Toxicology and Pharmacology: It's difficult, but important to make negative predictions. 

In addition, a recorded webinar by Richard Williams highlighting the science and functionality of Negative Predictions within Derek Nexus can also be viewed.

For more information on how to understand negative predictions in Derek, please view this infographic

 

General Approach

Lhasa Ames Test Reference Set

Performance Metrics for Bacterial Mutagenicity

Lhasa Skin Negative Predictions Dataset

Performance Metrics for Skin Sensitisation

 

General Approach

This functionality further evaluates those compounds which do not fire any bacterial in vitro (Ames test) mutagenicity alerts in Derek Nexus. The query compound is compared to a Lhasa reference set of Ames test data, producing the following outcomes:

  • In compounds where all features in the molecule are found in accurately classified compounds from the reference set, an Inactive prediction is displayed.
  • For those query compounds where features in the molecule are found in non-alerting mutagens in the Lhasa reference set, the prediction remains Inactive and the Misclassified1 features are highlighted to enable the negative prediction to be verified by expert assessment.
  • In cases where features in the molecule are not found in the Lhasa reference set, the prediction remains Inactive and the Unclassified2 features are highlighted to enable the negative prediction to be verified by expert assessment.

Derek Nexus also further evaluates compounds which do not fire an alert for skin sensitisation. The query compound is compared to the Lhasa skin sensitisation negative predictions dataset, producing the following outcomes:

  • Where all features in the molecule are found in accurately classified compounds from the dataset, a prediction of Non-Sensitiser is displayed. 
  • For those compounds where features in the query are found in non-alerting sensitisers in the Lhasa dataset, the prediction remains Non-Sensitiser, but Misclassified1 features are highlighted to enable the negative prediction to be verified by expert assessment.
  • In cases where features in the query are not found in the Lhasa dataset, the prediction remains Non-Sensitiser, but the Unclassified2 features are highlighted to enable the negative prediction to be verified by expert assessment. 

Please view the following infographics for more information on understanding negative predictions:

Understanding Negative Predictions in Derek Nexus

Understanding Negative Predictions in Derek Nexus – a detailed workflow

  1. Misclassified features are those that have been derived from non-alerting mutagens/skin sensitisers in the Lhasa reference sets.
  2. Unclassified features are those that have not been found in the Lhasa reference sets.

Lhasa Ames Test Reference Set

The Lhasa reference set of Ames test data is composed of a number of Ames test datasets, including, but not limited to:

Data SetNumber of CompoundsNumber of PositivesNumber of NegativesNumber of Equivocal or Inconclusive
Vitic NTP 2001 591 1235 175
Hansen 6512 3505 3008 0
FDA CFSAN 8421 4300 4115 6
ISSSTY 7363 3574 2665 1124
CGX* 718 360 343 15
Marketed Pharmaceuticals 554 40 494 20
Vitic Nexus Summary Call 6041 2905 3185  951 
Member Data 72  24  47 

 

All of the datasets were combined and curated (including the removal of duplicates, and compounds with equivocal, conflicting or inconclusive results) to produce the Lhasa Ames test references set with the following composition:

Data SetNumber of CompoundsNumber of PositivesNumber of NegativesNumber of Equivocal or Inconclusive
Lhasa Ames test reference set 9882 4716 5166 0

 

* CGX database can be downloaded from the Lhasa Limited CGX webpage

Performance Metrics for Bacterial Mutagenicity

Three proprietary datasets were used to assess the accuracy of the negative predictions; the results and composition of each can be seen below: 

Figure 1: The composition of the three datasets.

 

Vitic Intermediates data set No. of compounds Proportion in data set Negative predictivity

All compounds in analysis

1281

100%

n/c

Alerting compounds

582

45.40%

n/c

Non-alerting compounds

Inactive

664

51.80%

86.10%

Inactive with misclassified features

24

1.90%

66.70%

Inactive with unclassified features

11

0.90%

63.60%

Inactive with unclassified and misclassified features

0

0%

n/c

 

 

Proprietary data set 1 No. of compounds Proportion in data set Negative predictivity

All compounds in analysis

438

100%

n/c

Alerting compounds

51

11.60%

n/c

Non-alerting compounds

Inactive

334

76.30%

94.90%

Inactive with misclassified features

32

7.30%

84.40%

Inactive with unclassified features

21

4.80%

95.20%

Inactive with unclassified and misclassified features

0

0%

n/c

 

 

Proprietary data set 2 No. of compounds Proportion in data set Negative predictivity

All compounds in analysis

507

100%

n/c

Alerting compounds

86

17.00%

n/c

Non-alerting compounds

Inactive

383

75.50%

88.50%

Inactive with misclassified features

12

2.40%

50.00%

Inactive with unclassified features

26

5.10%

96.20%

Inactive with unclassified and misclassified features

0

0%

n/c

Back to top

Lhasa Skin Sensitisation Negative Predictions Dataset

The Lhasa Skin Sensitisation Negative Predictions Dataset is comprised of a mixture of human and animal data. In order to assign an experimental call for a particular reference compound, an assay hierarchy is used to rank the data. For example, human data is ranked above standard animal assays, which in turn is ranked above non-standard and other animal assays. A summary of the data can be seen in the table below.

Data source used to assign overall call Number of chemicals Proportion of overall datasets (%) Number of sensitisers Number of non-sensitiser Prevalence of non-sensitisers (%)

Human data

371

13

261

110

30

Standard Animal Assays

 2068

 75

898

1170

57

Non-Standard Animal Assays

302

11

162

140

46

Other Animal Assays

 25

1

25

0

0

 Total

2766

100

1346

1420

51

 

Performance Metrics for Skin Sensitisation

An external dataset of 986 compounds was used to measure the performance of the negative prediction functionality for skin sensitisation in Derek Nexus. The results from the external validation can be seen below in figure 1. 

 

Figure 1: The negative predictivity of each type of prediction during the external validation.

Misclassified and unclassified results are highlighted for expert review but are not necessarily an indication of activity.

The prevalence of each type of prediction can be seen below in figure 2.

 

Figure 2: The frequency of occurrence of each type of prediction during the external validation.

Back to top

© 2017 Lhasa Limited | Registered office: Granary Wharf House, 2 Canal Wharf, Leeds, LS11 5PS, UK Tel: +44 (0)113 394 6020
VAT number 396 8737 77 | Lhasa Limited is registered as a charity (290866)| Company Registration Number 01765239 (England and Wales).

QuestionPro supports sample survey questions such as multiple choice, drop-down menu, likert-scale, semantic differential, matrix, constant sum, drag-and-drop, slider-scale, net-promoter scale, and many more question types.