Can blockchain and in silico solve the reproducibility crisis?
With the world’s toxicological community gathering in Helsinki for Eurotox 2019 (8–12 September), now seems a good time to reflect on the growing challenge reproducibility poses to the chemical and pharmaceutical industries — and to science in general.
One of the cornerstones of science is the principle that claims and conclusions must be verifiable, i.e., that any other scientist following the same, precisely described steps should obtain (within a certain margin of error) the same results. Unfortunately, a recent study of 1500 scientists found that “more than 70% of researchers have tried and failed to reproduce another scientist’s experiments, and more than half have failed to reproduce their own experiments.”* This gap may grow as science becomes increasingly dependent on computational modelling — with its diversity of approaches, software tools, and hardware architectures — and as new methods emerge in the field of alternative testing to support safety assessment. Both industry and regulators are already finding the establishment of reproducible application, evaluation and guidance challenging. Regulatory acceptance is, as a result, slowing to a crawl. But maybe we already have a solution to the reproducibility crisis: by combining the right in silico toxicology (IST) protocols with a blockchain.
IST protocols
A scientific experiment is only as good as the data it generates and only when that data can be used and reused with confidence and allow the complete comparison of all steps, from raw data to results. IST protocols ensure that the digitization and entering of data, the documentation of data manipulation, the processing and, most importantly, complete in silicoapproaches (e.g., read-across and QSAR) follow high-quality data standards. So, errors during input and unintentional modification or disruption are avoided or, at least, minimized by standardized and consistent procedures with clear instructions.
Blockchain: the super-database
As the saying goes, a blockchain is a database but a database is not a blockchain. Indeed, whereas any database is subject to errors and changes, a blockchain provides an immutable, decentralized record of not only who did what when, but even how an experiment was carried out. For example, did the supplier of an active substance produce it according to the set standards? Or, did the distributors handle a temperature-sensitive product with sufficient care? So, a blockchain provides evidence that best practices in reproducible workflows have been respected. That means verification of source data can be done automatically, clinical trials can go faster and new products get to the market.
*http://www.nature.com/news/1-500-scientists-lift-the-lid-on-reproducibility-1.19970