How to capture data to share
As science-based companies strive to automate as much of their research and development as possible, enthusiasm for electronic laboratory notebooks (ELNs) is once again on the rise. The promise of capturing data about a drug, pesticide, or other chemical at the earliest moment in the product cycle, and then reusing it as widely as possible throughout the company, is an attractive one. So attractive that the idea has survived to make a comeback, after ELNs failed to make their mark when they were first introduced around 10 years ago.
The case for the new generation of ELNs was put recently at a conference in London organised by the International Quality & Productivity Center (IQPC) and sponsored by the sector's industry association, the Collaborative Electronic Notebook Systems Association (CENSA).
At the meeting, software suppliers including Waters and Intellichem set out the benefits of ELNs. Thorsten Froelich, director of Waters Informatics, described their potential role in data integration, quoting a dictum of Professor Rolf Krebs, the former chairman of Boehringer Ingelheim: 'Data integration is the key priority for pharmaceuticals and biotechnology vendors today.' The deployment of ELNs would, he said: increase research and manufacturing accuracy; increase the speed and efficiency of R&D; improve data management; improve prediction and interpretation of research data; help transform hits (potential drugs) into leads; and more.
- IC Vault provides access to document-management capabilities, including security settings and in-depth searching.
Intellichem's Matt Wallach, too, stressed the way data integrated under ELNs would improve corporate strategy by allowing information to be used more intelligently with the help of decision-support software - for example, by improving technology transfer between drug discovery and drug development departments.
These are just two of around two dozen software suppliers with ELN solutions of one sort or another. Vendors range from veterans, such as Intellichem and Cambridgesoft, to relative newcomers to the field, such as Waters and Tripos. Products are broadly split into two camps - discipline-specific and universal offerings. Universal or generic software provides an architecture and tools to create and search content, and to work collaboratively, that will satisfy the needs of almost any science-related industry - Waters' product falls into this category. Discipline-specific ELNs are aimed at a particular market segment - typically the life sciences sector. MDL and Tripos, for example, have developed highly tailored systems based on their other chemistry software tools. There is, of course, overlap between the categories. Cambridgesoft and IntelliChem, who both started out supplying chemistry software, have more broad-based ELN offerings than their historical markets suggest.
One indicator of the renewed interest in ELNs, is the recent acquisition of Intellichem by Symyx. Intellichem leads the sector, with installations at five of the world's top 10 pharmaceutical companies: Eli Lilly; Bristol-Myers Squibb; Merck; Pfizer; and AstraZeneca. The most recent of these - at AstraZeneca's process chemistry division - is one of the most ambitious, in that it eliminates completely the use of paper notebooks. But each supplier has its strengths and weaknesses. For example, Cambridgesoft, with its heritage in desktop software, probably appeals to companies with fewer scientists. How does a life-science manager choose between them? And, more importantly, how does one avoid the pitfalls? After all, if they are so useful, why did ELNs fail to make much ground the first time round? As individuals in a similar position - when looking for a plumber, for example - we would probably ask around our friends and colleagues. But pharma executives can hardly hang around the coffee machine chatting to their counterparts at a rival firm. Instead they go to hear them at conferences like the one organised by IQPC.
Here, they can not only see many of the major software suppliers in one place, but can hear from industry consultants and companies wrestling with brand new systems. Proof of the resurgence in electronic laboratory notebooks is that pharmaceutical sector managers appear to be going to these events in increasing numbers. Greeting delegates in London, Colin Sandercock, of Heller Ehrman White & McAuliffe, a veteran of ELN conferences said that, after 10 years of attending CENSA-sponsored meetings with 20 to 50 participants, he was pleased to be standing in front of 150 that day in September.
- IntelliChem Discovery includes an intuitive one-page editor tailored for medicinal chemists.
CENSA chairman Richard Lysakowski had been billed to welcome delegates with an overview of the sector. In his absence, Lysakowski's talk was given by Simon Coles of Amphora Research Systems. For the ELN customers in life sciences, the stakes are high as they try to profit from the synergies between different parts of their businesses, he said. The history of ELNs is a troubled one, he warned, and prompts such questions as 'Will it be used?' and 'Can we afford it?' Yet the case for moving away from paper is compelling, as companies strive to get the right products to market faster.
CENSA was set up to develop a consensus on ELN systems but, looking at what life-science companies want from electronic R&D pipeline management software, suggests that there is no single solution. This drive to collaborative eR&D is a 'lofty goal'. 'It is structurally impossible to find one system that does it all,' hence, companies should look to adopt open formats and then integrate data, said Coles. This was essential, given the long development cycle and patent life of drugs. Companies have to ask whether a vendor of a closed system would still be around if a patent challenge occurs, when intellectual property has a life span of 40 to 100 years. The IT industry is immature when it comes to the understanding of archiving records. The great challenge is to design for preservation, said Coles. 'It is much larger than the Y2K problem, and will necessitate substantial changes if the target is to be met.' This didn't mean only file formats, but also how to store records on disk together with users' credentials. It has to be remembered that human resources departments can expire employee records after five years because of data protection requirements, he said.
There are plenty of obstacles to success. Inadequate knowledge of products and suppliers can be overcome. And very few organisations would be able to put an ELN in one go, said Coles, so they will need to set intermediate goals for their adopted software system.
Gaining the wholehearted support of senior management can be problematic, Coles suggested. Managers often ask for ROI (return on investment) projections to justify a project, and producing these can be 'very tricky'. Once gained, the support needs to be sustained if the ELN deployment is to be successful. A company with a workforce under 200 to 400 shouldn't undertake an ELN deployment, Coles suggested, because such a project needed a dedicated manager and an in-house IT department.
- IntelliChem Synthesis enables scientists to easily create structured procedures.
Legal and regulatory uncertainty about using ELNs to back up patent claims is not a sufficient justification for ignoring them, Coles emphasised. Although there has yet to be a legal test-case of the issue, the risks of moving to electronic records are insignificant compared to loss of intellectual property under the current regime. Significant numbers of paper notebooks escape proper witness signing every six months, he said. This can render them useless for patent justification.
Delegates listened to the bold claims of the vendors, and cautious optimism of the consultants, but were keenest to hear the talks by those had already embarked on (or completed) installation of an ELN - as judged by the number of questions they asked. And despite the advice offered by Simon Coles, several of these were smaller companies content to defy the common wisdom that an ELN is too much trouble, too expensive, and unnecessary for firms with fewer than 250 staff.
For example, Hongmei Huang, senior manager in charge of data analysis at Colorado-based drug discovery firm Array Pharma, described how the company set out to 'enhance communication' between its 180 scientists using an ELN. Five years ago, the company had a fairly modest goal - to have a searchable database to share expertise amongst chemists only. Over the years this grew into a standardised notebook, integrating analytical and experimental information with systematic tracking of compliance of notebook-signing need for intellectual property protection.
Because of its size, Array chose to buy-in ELN software rather than develop its own system. 'In 1999 there were not many options,' said Huang, with Cambridgesoft being chosen despite its product not then being mature. There was a degree of risk to the strategy, but benefits too. The 'growing pains' of being an 'early adopter' were compensated for by Array's influence on the development of the software. Getting people to use the software was always going to be a 'big issue' and a key to the success of the project. So the software needed to 'to do what scientists want.' And because the ELN was going to be used 'all day long' by everyone, performance was an issue. Crucial then, to the project, was the programme to try out early versions on a selected 'focus group' of chemists, she said. With the feedback from multiple rounds of testing, and liaison with Cambridgesoft, there was a steady incremental improvement in the system.
The focus group chemists then trained their fellow scientists, which helped convince the sceptics in the organisation. The roll out was carried out a department at a time. Each chemist now has a computer in their office, and in the laboratories two people share a terminal.
The advantages are clear. The biggest 'has been the protection of [Array's] intellectual property,' said Huang - although it was not one of the initial goals of the project. The method of printing out every 100 experiments for witness signing is a simple and efficient process. Time is saved in data entry and in patent preparation. Expertise is more easily shared among scientists who, 'are writing better procedures, because other people see them,' said Huang.
As the system stands, users are now very happy with it and its use by chemists has become mandatory. But Array has become a victim of its own success, said Huang. The system is currently moving from a client-server to a web-based version of the software. There are other opportunities for improvement, including better search capabilities, greater inclusion of analytical data, integration with the chemical inventory system, and wider deployment within the company (to biologists and process chemists). However, there is resistance to change. Having learnt one system, scientists are reluctant to switch to a new one, especially one that might 'break' existing ways of working. In addition, a significant change to the database design will involve migration of a sizeable amount of legacy data.
To overcome this, Array will follow its previous practice of consultation and education, involving users in early testing, and group training before a major software rollout, with supplementary training after upgrading, said Huang. Having started with a modest goal, Array is now on the way to a fully-fledged ELN. But even Array prints out its notebooks rather than adopting digital signatures. As to the one remaining paper step - 'let someone else be the pioneer,' says Huang.
Some of the same lessons emerged at AstraZeneca's process chemistry group, where Intellichem's ELN was the chosen system. Although many of the advantages from ELN adoption are corporate ones - such as IP protection - there also needed to be benefits for individual scientists - such as information sharing and time saving - and these needed to be communicated effectively to the staff.
John Jones, AstraZeneca's global project manager for the Intellichem deployment, explained how a small group of chemists was trained to use the software. Enthusiasm generated from these participants then helped build support for wider deployment of the ELN, which was carried out in stages.
There were differences, too, from Array's experience. Because the ELN had to be used on multiple sites, common guidelines had to be agreed on working practices that would be codified in the system. 'Large global meetings were needed to agree a way forward,' said Jones. All of these meetings and staged rollouts took a long time - two to three years. As to the benefits: 'Experiments are more legible and follow more consistent patterns,' Jones said, and he was 'quite astonished' by the efficiency gains from reusing experimental data. Scientists reused 60 to 70 per cent of their own work, and there was 25 per cent re-use of experimental data produced by other scientists.
Delegates also heard about the implementation of Tripos' ELN at Schering sites in Europe and the US (see June 2004 issue of Scientific Computing World). And experts from notebook suppliers Synthematix, Klee Group, and Velquest were also on hand to field questions from delegates.
On the evidence of the conference, it's a good time to invest in an ELN - products are maturing and competition is intensifying. But the real lesson of those who pioneered their introduction: pay attention to the human dimension if you want your implementation to succeed.