Scientific research is changing. In order to increase the time to market, researchers want to break down the silos between departments, simplify the way they work and focus on doing science.
However, their biggest frustration (especially in chemical research) is that they know they are repeating reactions which may have been done countless times before. With paper-based lab books, there is no way to know in advance that a particular synthesis is doomed to failure. Unless the researcher can speak to a colleague who has tried the same reaction before, there’s no way of finding out. And when the reaction fails, regrettably, they don’t tell anyone about it, thus perpetuating the cycle.
So, why don’t researchers share failed experiments?
The publishing conundrum
All researchers start in academia, and adopt a working practice of reading peer-reviewed journals as a source of information. However, journals don’t accept articles for unsuccessful reactions and funding bodies are unlikely to reward someone with a track record of failure. Even well-conducted and adequately executed studies may be rejected by journals if the results are not deemed to be exciting. Moreover, many universities have restricted library budgets, meaning that they may only subscribe to the better-respected journals, further limiting the available information.
There is also an assumption that the information in publications is correct. Indeed, the Economist published an article in 2013, suggesting that “…there are errors in a lot more of the scientific papers being published, written about and acted on than anyone would normally suppose, or like to think.”
Many researchers don't even consider trying to publish failed experiments. So, whilst research advances knowledge overall, only a fraction of this is accessible in the form of publications. Some sources say only 5% of scientific work gets published, and the rest of it sits in paper or electronic repositories, never to see the light of day. Consequently, there are countless labs wasting their time doing experiments that have been tried hundreds of times before, and shown not to work. The community is held back because this information is not shared, and this ethos of not sharing failed experiments is taken forward into industry.
Recently, there have been efforts to try to break this cycle. The ChemSpider Synthetic Pages provide a free database of practical procedures for synthetic chemistry, which is written (and validated) by chemists, for chemists. This database provides insights to frequently encountered problems, repeatability and scalability of reactions. However, most of the contributions are from academia, presumably because large pharma are concerned about IP and retaining their competitive position.
Organizational culture plays an important role. Assuming that researchers have overcome their academic reticence to share data, there may be further philosophies to negotiate. Knowledge may be seen as a source of power within the organization, and chemists may be reluctant to impart their knowledge in the belief that it will weaken their competitive position. Other cultural factors may come in to play, with researchers being reluctant to share data in the belief that it will bring criticism of their work, or be used to unfairly assess their performance.
Assuming that the cultural barriers have been overcome and that researchers are willing to share data, the next challenge is actually finding it. The format in which the research data is stored may include paper printouts of spectra, associated instrument data files and spreadsheet documents used in conjunction with experimental write-ups. These paper and file-based systems mean that data is constrained to a locale, and may be inaccessible to researchers at another site.
Using contract research organizations (CROs) to fulfil requests for intermediate or final compound synthesis is commonplace. Whilst their services mean that compounds can be made faster or cheaper, what happens to the knowledge and insights that are gained by the CRO during the investigation of synthetic routes? Is this valuable information also communicated to the requesting organization, or is it just a report of the successful synthesis route? It may be that CROs themselves are constrained under the terms of agreement and may not publish anything related to commercial projects, and this again contributes to the death of data.
Failure is good
At IDBS, we are building the next-generation of chemistry electronic laboratory notebook (ELN) to facilitate the capture of research data using an intuitive web-based platform for the exchange of information. Using an ELN to capture chemical reactions and supporting data, allows chemists to have greater visibility of the overall research effort (and it should be argued that this be across the entire research organization). IDBS also recognize that CROs are routinely used to outsource synthesis, and have developed E-WorkBook Connect, a cloud-based collaboration platform to facilitate communication and tasks with external agents. As IDBS progress with the development of the products, they will merge, allowing the organization to retrieve all experimental data from the CRO into the ELN, as if the research were being conducted in-house.
Sharing data (and having complete visibility across the organization) is the way forward. Chemists need to learn from each other’s mistakes, not only their successes. In doing so, the amount of time, energy and costly experiments can be reduced, helping chemists reach their desired end goal faster. Free access to public domain databases will help to facilitate this, but it requires active collaboration from both industry and academia.
The entire research process is fundamentally focused on success, and this needs to change.