Analysis of the benefits of imputation models over traditional QSAR models for toxicity prediction
Abstract Recently, imputation techniques have been adapted to predict activity values among sparse bioactivity matrices, showing improvements in predictive performance over traditional QSAR models. These models are able to use experimental activity values for auxiliary assays when predicting the act...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2022-06-01
|
| Series: | Journal of Cheminformatics |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s13321-022-00611-w |
| _version_ | 1818253025854095360 |
|---|---|
| author | Moritz Walter Luke N. Allen Antonio de la Vega de León Samuel J. Webb Valerie J. Gillet |
| author_facet | Moritz Walter Luke N. Allen Antonio de la Vega de León Samuel J. Webb Valerie J. Gillet |
| author_sort | Moritz Walter |
| collection | DOAJ |
| description | Abstract Recently, imputation techniques have been adapted to predict activity values among sparse bioactivity matrices, showing improvements in predictive performance over traditional QSAR models. These models are able to use experimental activity values for auxiliary assays when predicting the activity of a test compound on a specific assay. In this study, we tested three different multi-task imputation techniques on three classification-based toxicity datasets: two of small scale (12 assays each) and one large scale with 417 assays. Moreover, we analyzed in detail the improvements shown by the imputation models. We found that test compounds that were dissimilar to training compounds, as well as test compounds with a large number of experimental values for other assays, showed the largest improvements. We also investigated the impact of sparsity on the improvements seen as well as the relatedness of the assays being considered. Our results show that even a small amount of additional information can provide imputation methods with a strong boost in predictive performance over traditional single task and multi-task predictive models. |
| first_indexed | 2024-12-12T16:33:31Z |
| format | Article |
| id | doaj.art-9c394aa713b1463b9f35b8b720813f4f |
| institution | Directory Open Access Journal |
| issn | 1758-2946 |
| language | English |
| last_indexed | 2024-12-12T16:33:31Z |
| publishDate | 2022-06-01 |
| publisher | BMC |
| record_format | Article |
| series | Journal of Cheminformatics |
| spelling | doaj.art-9c394aa713b1463b9f35b8b720813f4f2022-12-22T00:18:43ZengBMCJournal of Cheminformatics1758-29462022-06-0114112710.1186/s13321-022-00611-wAnalysis of the benefits of imputation models over traditional QSAR models for toxicity predictionMoritz Walter0Luke N. Allen1Antonio de la Vega de León2Samuel J. Webb3Valerie J. Gillet4Information School, University of SheffieldInformation School, University of SheffieldInformation School, University of SheffieldLhasa LimitedInformation School, University of SheffieldAbstract Recently, imputation techniques have been adapted to predict activity values among sparse bioactivity matrices, showing improvements in predictive performance over traditional QSAR models. These models are able to use experimental activity values for auxiliary assays when predicting the activity of a test compound on a specific assay. In this study, we tested three different multi-task imputation techniques on three classification-based toxicity datasets: two of small scale (12 assays each) and one large scale with 417 assays. Moreover, we analyzed in detail the improvements shown by the imputation models. We found that test compounds that were dissimilar to training compounds, as well as test compounds with a large number of experimental values for other assays, showed the largest improvements. We also investigated the impact of sparsity on the improvements seen as well as the relatedness of the assays being considered. Our results show that even a small amount of additional information can provide imputation methods with a strong boost in predictive performance over traditional single task and multi-task predictive models.https://doi.org/10.1186/s13321-022-00611-wQSARImputation modelingMulti-task modelingToxicity predictionModel evaluation |
| spellingShingle | Moritz Walter Luke N. Allen Antonio de la Vega de León Samuel J. Webb Valerie J. Gillet Analysis of the benefits of imputation models over traditional QSAR models for toxicity prediction Journal of Cheminformatics QSAR Imputation modeling Multi-task modeling Toxicity prediction Model evaluation |
| title | Analysis of the benefits of imputation models over traditional QSAR models for toxicity prediction |
| title_full | Analysis of the benefits of imputation models over traditional QSAR models for toxicity prediction |
| title_fullStr | Analysis of the benefits of imputation models over traditional QSAR models for toxicity prediction |
| title_full_unstemmed | Analysis of the benefits of imputation models over traditional QSAR models for toxicity prediction |
| title_short | Analysis of the benefits of imputation models over traditional QSAR models for toxicity prediction |
| title_sort | analysis of the benefits of imputation models over traditional qsar models for toxicity prediction |
| topic | QSAR Imputation modeling Multi-task modeling Toxicity prediction Model evaluation |
| url | https://doi.org/10.1186/s13321-022-00611-w |
| work_keys_str_mv | AT moritzwalter analysisofthebenefitsofimputationmodelsovertraditionalqsarmodelsfortoxicityprediction AT lukenallen analysisofthebenefitsofimputationmodelsovertraditionalqsarmodelsfortoxicityprediction AT antoniodelavegadeleon analysisofthebenefitsofimputationmodelsovertraditionalqsarmodelsfortoxicityprediction AT samueljwebb analysisofthebenefitsofimputationmodelsovertraditionalqsarmodelsfortoxicityprediction AT valeriejgillet analysisofthebenefitsofimputationmodelsovertraditionalqsarmodelsfortoxicityprediction |