An ELN by any other name
Electronic laboratory notebooks (ELNs) represent one of the fastest-growing informatics markets, with a growth rate of more than 30 per cent a year, and double-digit growth will continue over the next four years, according to a recent survey from Atrium Research (see page 53).
Despite – or perhaps because of this growth, with more than 30 suppliers now offering commercial solutions – some widely divergent views on the purpose and function of ELNS were on display at the ‘ELNs and Laboratory Informatics 2006’ conference, in London at the end of September. On two topics, however, there was widespread (though not universal) agreement – that a generic ELN is a chimera; and that ELNs should be regarded as much as a medium for communication as for record-keeping.
One sign of the huge interest in ELNs was the launch at the conference of an ELN developed by a company that specialises not in laboratory informatics at all but in product lifecycle management (PLM) software. The Texas-based UGS Corporation developed its ‘Research Knowledge Management’ suite of software in collaboration with Procter and Gamble’s global research laboratories. The RKM product will be released as a core module in UGS’s Teamcenter product lifecycle management software.
ELNs for chemistry dominated the conference, though exhibitors included IDBS, whose ‘BioBook’ specialises in the life sciences. It also opened up the big question of preservation of digital records, with a variety of views as to whether one should store simply the notebook recording the experiment and results, or whether, as well, one should try to archive the raw data for future data-mining.
Simon Coles, chief executive of Amphora Research Systems, stressed the point that different scientists do different sorts of research in different disciplines: ‘Anyone who claims one solution solves all problems is... a salesman.’ The point was also made by David Dorsett, vice-president and general manager of Symyx, who said that his company’s ELNs had ‘specific features for specific communities. We have to have something that speaks their language. Our product philosophy is specific ELNs rather than enterprise-wide, generic ELNs.’
Although much of the early promise of ELNs was that they would capture knowledge in the research laboratory for re-use elsewhere in the enterprise, some of the most successful installations appear to be at the process stage. Dr Michael Kopach, principal research scientist at Eli Lilly, described how, in the company’s four-year implementation programme for ELNs, it decided to deploy first in process before including the research and discovery stages. Dr Dorsett noted that discovery chemistry had had informatics support from MDL and Accelrys for years, whereas ‘process was not adequately served by the cheminformatics industry’.
Coles, too, addressed this theme. ‘Chemists are the most vocal,’ he said. ‘Organic chemistry has a clear, widely-used notation, which rewards specific support. And there is a cheminformatics industry. But other disciplines are much more specialised. Chemistry-centric ELNS don’t tend to be well received outside of chemistry. Chemistry tends to be pretty structured and routine, whereas other disciplines have much greater variation in their work.’ Companies do not need to buy one system, he said, because integration of different software is becoming a fact of life.
One of the factors favouring ELNs is quite simply that the younger generation of scientists is leaving university with the expectation that they will use computers routinely in their work: ‘They are more used to PCs than paper,’ Dr Dorsett said. But Symyx still has to show ‘hard’ ROIs (returns on investment) to convince its customers to spend the money on an electronic system: ‘The business side of things is now driving adoption.’ The promise that researchers would avoid needlessly repeating earlier experiments – simply because they did not know that such experiments had already been carried out – has been one of the highest expectations of ELNs. But, perhaps surprisingly, Dr Dorsett believes that this is really only a ‘soft’ ROI: ‘You should see less rework, but it is impossible to substantiate this,’ he said.
To understand the ‘hard’ ROI, he suggested, one had to look at what scientists actually do during their day’s work. Do they leaf through odd bits of paper to find things? Are they trying to find data to write up their conclusions? Are they having meetings with the patent people to try to explain what they did six months ago? It is the concept of a notebook as a means of communication – a portal – that he believes is emerging as one of its most important aspects. ‘The legibility of ELNs is better for patent purposes than paper notebooks,’ he said, ‘and this is a tremendous benefit to companies. It’s the accessibility of information to the patent person who dives in to organise the patenting, without the context of the project team. Also the benefits for someone coming in to shift [the project] to manufacturing. All these pressures are the same in terms of communications issues.’ Good communication is especially important, he noted, when pharmaceutical companies outsource work to contract research organisations (CROs). Using an ELN, CROs can more rapidly deliver results to their principals in formats that are better and more secure than PowerPoint.
The deployment of an ELN requires a continuing commitment by management and ‘effective advertising’ to the potential users, as Dr Kopach’s description of users in the Eli Lilly project illustrated. It took the decision to deploy to 100 users to begin with and collect at least six months’ worth of operational data before trying to roll it out to more users. The company is now deploying the system among its CROs.
The project needed support from the legal department, QA/QC, and from IT, he noted. For the lawyers, the big item was electronic signatures of the lab notebooks, so the company instituted a business rule that the scientist should sign electronically on a daily basis and that there should be a counter signature within 30 days of the creation of a new record. Compliance with this rule is currently at about 80 per cent – however, an audit of paper lab notebooks revealed only a 50 per cent compliance rate, so the performance of ELNs is better than paper.
The attention to detail by management to enhance user acceptance extended to the details of the design and location of the hardware. The PC was put on the top shelf of the laboratory, away from the ‘splash zone’ where it might be damaged by a chemical spillage. The flat-panel monitor was mounted on a swing arm to keep bench-space clear, with a wireless connection between the keyboard, mouse and the PC itself. The PC is hardwired into the company network, not via a wireless connection. Only the final version of the ELN is archived, in pdf format.
So far, user-surveys have not found that people are searching the ELNs for information. ‘We need to improve on that,’ Dr Kopach said. However, the company did find that the notebooks were being ‘cloned’ – users were copying experiments from one to the other. This was, he said, one of the biggest value-adds of the project.
The theme of ELNs as a format for communication was prominent in the presentation by Dr Ulf Fuchslueger, chief executive of the Swiss company Vialis. He stressed that ELNs should be neither an isolated silo of information nor an island of automation for a specific group of users. Instead, he maintained, they should ‘provide the required information to the scientist’. He suggested that further benefits of ELNs will come when they are integrated with a company’s enterprise resource planning (ERP) software, but he also echoed Dr Kopach’s point of the need for commitment by higher management to the deployment and integration of the different software systems. ‘ELN/ERP integration is a political issue and it’s hard to get [higher-level management] people to understand lab requirements,’ he said.
He cited two case studies – one of them in a small CRO – where the implementation of an ELN that was integrated with the company’s ERP reduced the time taken to prepare a quotation to a customer from two weeks to two days, with benefits of more consistent and lower pricing and making the preparation of quotations easier.
But another important aspect of implementation, highlighted by John Trigg from phaseFour Informatics, is to avoid conflicts between the laboratory and IT department. ‘The situation now is all but intolerable,’ he said. ‘No matter who you are, unless you are the CTO, the IT department does not work for you, so your interests, and their interests, are not aligned.’
Dr Trigg noted that the two departments work to different criteria: the IT department is employed to maintain stability, whereas scientists have to take risks if they are to make discoveries. The IT department is driven by ‘total cost of ownership’, standardisation and a one-size-fits all approach, whereas scientists are interested in long-tail applications.
Trigg’s solution involves greater communication between the two parties. He advises developing a problem statement that specifies the issue in measurable terms; defining achievable corrective actions, such as open house sessions and hardware amnesties; and constant review of progress.
The application of this to ELNs is, he believes, that ‘IT needs to be fully involved from the start, and have a big say in the technology used. They may already have the skills relating to electronic records management and integration with other corporate systems.’
Michael Speed from Organon, which uses an ELN from CambridgeSoft, had a similar view. He said: ‘We created a communication plan, and organised a teleconference with CambridgeSoft. Face-to-face meetings were essential.’ In addition, he believed it was important not to aim for perfection, and to accept you can’t integrate everything.
Cat Tarlton, senior scientific data analyst at Array Biopharma, believed in a less sympathetic, ‘stick-and-carrot approach’, to gain acceptance from the laboratories. ‘For Array, the strategy must be driven by high-level business needs, while adoption must be driven by tactical advantages for the end user, to empower every scientist to make better decisions and increase productivity.’
The ‘carrots’ to lead the scientists to acceptance of the change were: to give the ELN the look and feel of a paper notebook and to create integrations between key applications. Another benefit was the ELN was programmed for automatic integration of data from lab instrumentation, including a voice recognition system for data entry from microscopes, and a digital camera station for biologists. These all contributed to ‘making an ELN that’s so easy, your grandmother could use it.’
These were then followed up with the ‘stick’: higher management demanded that they made the change, and no more hardbound notebooks were available. The main objections to an ELN were that computers crash and could lose data, and ‘Big Brother Syndrome’ – a fear that everyone, including the boss, could access someone’s files and check they were on track.
The process of signing and countersigning to protect intellectual property rights was a big concern for lawyers from the outset of the ELN. Originally, Array used a pen and paper ‘cut and paste’ method, which was inefficient and time consuming. It has since moved to e-signatures, using PatentSafe, which is fully automated and saves cost.
Tarlton also took up the theme that one size does not fit all. Although Array is using a generic notebook from CambridgeSoft, it has customised the software for each discipline, using discipline-specific templates designed for ease of use. Biologists particularly had had initial problems, due to the variety of their workload, and the huge amounts of data they have to process. The discipline-specific modification has produced a format that is particularly suitable for biologists, allowing integration of pictures with annotations.
Although much of the attention at the conference focused on developing and implementing ELNs, there was also considerable debate about archiving and about data storage. Partly, this is driven by the hope that, as science develops, researchers might want to ask different questions of data that had already been gathered for quite a different purpose. Rather than repeating the experiment, so the hope goes, researchers might be able to deploy advanced data-mining tools to interrogate repositories of data and yield useful new knowledge. Partly also, it is driven by fear – that a patent claim might be challenged at some point in the future or that a drug may have to be recalled due to an adverse drug reaction and the pharmaceutical company would then have to produce the data to defend its intellectual property or to show that it had properly considered and tested drug toxicity issues.
Dr Dorsett stressed that Symyx does not view ELNs as a repository for raw data. This is not an issue of computational capacity, he said: ‘it’s two different systems. The notebook documents what people wanted to do; what they did; and what they learned.’ However, Dr Robert Scoffin, of CambridgeSoft, told the conference that some of his customers had opted to use the CambridgeSoft ELNs as data stores as well. Simon Coles, from Amphora Research Systems, was categorical: ‘An ELN isn’t the place for raw data. It can provide metadata but store the raw data elsewhere.’ Private conversations in the corridors evinced a mixed reaction, with some radicals advocating that raw data should be deleted after the experiment was completed, so that the only legally admissible record was the laboratory notebook’s account of the conclusions of the experiment. If the raw data no longer existed, this eliminated the risk that it might be re-interpreted at a later date during legal proceedings. However, GlaxoSmithKline appears to be following the route of preserving the raw data, as outlined by Mohan Cashyap, assistant director for pre-clinical IT at the company. A particular problem that has driven the company down this route is that even data taken within the past eight years is close to unreadable because of obsolescent laboratory hardware, so the company needs to preserve and update its records just to ensure that current projects are fully documented.
Tony Davies, formerly a senior informatics marketing manager with Waters and now external professor at the University of Glamorgan in the UK, warned the conference that while the ELN market was maturing, there was still an immature marketplace when it came to the technology for preservation of raw data. There had been a plethora of mergers, spin-offs and take-overs among the instrument vendors and associated scientific document management system providers. Examples included the acquisition of NuGenesis and Creon Lab control by Waters and of SSI by Agilent. Inevitably, this raises questions about how long some products will be supported.
Other hazards, he noted frankly, included vendors’ lack of understanding of customers’ needs coupled with customers’ lack of understanding of the complexity of the problems/workflows; widespread misunderstanding of language and terminologies in use; and poor communication between vendors and customers. Success rates had been relatively poor, he said, with legacy data because there had been little user buy-in, little management buy-in, and the projects had been driven by the decommissioning of the systems that generated the legacy data in the first place. In contrast, there had been good success rates with current, productive data because management could see the need for regulatory compliance and also how archiving data would enhance IP protection for the drug development pipeline. He concluded with a warning that archiving should be built in and not added on as an afterthought and that, as customers’ expectations increased, no vendor of analytical instrumentation would be safe without offering this sort of future-proofing.