Effiris delivers state-of-the-art secondary pharmacology models, where each model is able to learn from the private data of all pharmaceutical collaborators. As a result, Effiris overcomes one of the main challenges facing model development in drug discovery – accessing and utilising high-quality proprietary data.
Prior to this release, Effiris already offered model building – the ability to build teacher models in Effiris using in-house confidential data, privacy-preserving data sharing – the production of label files which extract knowledge while preserving all confidential aspects of the compounds used, and hybrid model building – collating Effiris member’s confidential dataset labels with the shared knowledge available in Effiris and publicly available data, in order to generate more comprehensive dataset to build a hybrid models*.
What can you expect from Effiris 2.2?
- 49 supported secondary pharmacology endpoints. An additional 32 relevant secondary pharmacology endpoints since our last release. This list was extended using knowledge of which endpoints are most used within the pharmaceutical industry – guided by Lhasa research and the Effiris consortium.
- Choice over which datasets to use. When building hybrid models in Effiris 2.2, there is now choice over which available datasets to use for each type of data.
- Model validation. External validation can be run against both teacher and hybrid models using a validation dataset. Validation results are presented in a new, user-friendly format that is easy to view, understand and export. The option to run internal 5-fold cross validation against built hybrid models has also been included.
- New dedicated predictions page. With the option to select which models to use when running predictions.
- Enhanced data upload. Effiris now supports data upload for an increased number of file formats and the upper limit of structures per member dataset has increased.
Effiris 2.2 is available now. If you would like to minimise data gaps in secondary pharmacology, contact us to find out more about Effiris.
*The benefit of the Effiris hybrid model approach is shown in this infographic ‘Anticipating and mitigating adverse drug reactions through machine learning and privacy-preserving data sharing’.