July 1, 2007 (Vol. 27, No. 13)
To Meet Evolving Needs, Users Cite Flexibility as the Most Important Element for an ELN
Electronic lab notebooks (ELNs), once the purview mainly of synthetic and medicinal chemists, are expanding in terms of applications and business value to become data portals and consolidation points for experiments throughout the enterprise. In the process, they are providing scientific context for a wide range of data, transforming it from information to knowledge, according to speakers at the recent “IQPC” meeting in Frankfurt.
Today’s ELNs are being rolled out for enzymology, molecular and cellular biology, pharmacology, safety, toxicology, pharmacokinetics and pharmacodynamics, as well as for the omics fields, according to Paul Denny-Gouldson, Ph.D., product manager, IDBS (www.idbs.com).
The ELN requirements to support these fields are much more complicated than those for chemistry applications, Dr. Denny-Gouldson said. They require tools for data federation, data consolidation, statistical analysis, and data visualization, creating a gap between ELNs designed for chemistry and those for biology or generic applications.
As Dr. Denny-Gouldson noted, “While generic or chemistry-focused ELNs may address intellectual property requirements, they lack dedicated tools for biologists and typically only provide a cut-and-paste repository for data, creating another step in data-processing workflow and another data silo with limited structure content.”
Because biology experiments are conducted over time, often by many people, the system needs the flexibility to respond in real-time to mid-experiment changes. And, Dr. Denny-Gouldson emphasized, biology experiments tend to be “much more subjective than the more empirical sciences,” and thus require a more dynamic approach to experimental design and data capture. The results, too, “tend to be context-rich descriptions of the outcome” and may only be valid for a specific population set.
To address these evolving needs, flexibility is the most important aspect of any ELN, according to Markus C. Hemmer, Ph.D., senior product manger, Waters (www.waters.com). Simplifying quality control is a focus of Waters’ ELN work. As Dr. Hemmer elaborates, interoperability allows ELNs working with LIMS to dynamically populate documents, eliminating repetitious steps.
Their use also is adding value by “improving the authoring of experimental records, providing better access to records, and making the signing and archiving of records faster and easier,” noted Carl Meinhof, Ph.D., principal business systems analyst, IS, R&D informatics, Amgen (www.amgen.com). Specifically, he said, ELNs eliminate the need to “cut and paste with glue and scissors,” and make it easier to locate relevant documents, ensure they contain consistent information, and, importantly, eliminate the need to decipher handwriting.
Thomas Rozlucki, vp, product development, Contur Technologies (www.contur.se), reported that Contur’s clients are seeing overall gains through having information readily available. Initially, clients are focused on scientific productivity, “but, after a few years, they realize the value of a central repository.”
“The return on investment,” Dr. Denny-Gouldson said, “doesn’t come from end-users alone, but from a combination of end-user, IT, and business benefits.” These are being demonstrated in large, global projects, and in small, localized pilots that are completed in one month.
One client, Rozlucki said, developed an ELN business case that documented annual time savings per scientist at 20 days per year. Lonza (www.lonza.com) figures a return on investment at “E15,000 per year at least, each time a user uses the ELN to access LIMS data,” according to Matthieu Giraud, Ph.D., project manager, R&D, Lonza.
The ability to create templates for documents and for specific sections within documents is making ELNs increasingly functional. Most ELNs use document templates to provide context to raw laboratory data. Predefinable electronic sections help users organize data in terms of workflow. “You can enter data on the system as well as incorporate business logic from simple calculations to complex rules derived from standard operating procedures,” Dr. Hemmer said.
Linking between LIMS and ELNs, he said, allows seamless integration between scientific documentation and workflow. As an example, an order may begin with the LIMS and go to the ELN system, creating the required test documents dynamically, based upon established rules. The result is improved regulatory documentation, reduced repetition, and better control of the release of information.
Lonza launched a pilot ELN program to develop faster, more efficient access to R&D information for increases in intra-departmental collaborations throughout the company. The implementation began with its fine chemical division and is expanding enterprise-wide to include its synthesis and biopharmaceuticals teams and its bioscience development teams in Europe, North America, and Asia.
A second part of Lonza’s project, linking the ELN to the existing LIMS and exporting a product sheet from a specific experiment, have been completed since the meeting. The next step is to add a chromatography data system.
The resulting system, developed with Sopheon (www.sopheon.com), is part of an integrated information network that revolves around a chemical information system that contains substance information and links to the documents involving those substances. Within that framework, the ELN constitutes the reaction database.
The pilot’s hybrid approach, outlined by Dr. Giraud, uses scanned documents as well as online generation, so that all experiments—even incomplete ones—are accessible globally. And, to leverage the work, the experiments were copied as templates for future experiments. For reporting, data was exported to Excel. The system can import many different types of data, including experimental protocols, analysis, drawings, literature reviews, certificates of analyses, and release sheets.
State of the Industry
Right now, ELN means different things to different people, and the various systems have been developed in different ways—either from existing LIMS or from the ground up—and from different application areas. With such disparities, it’s no wonder there are problems.
“In QC, you know the outcome. In R&D, you make experiments and proceed step-by-step,” Dr. Hemmer said. So, ELNs designed from a quality-control perspective tend to be less flexible and have more restrictive handling requirements than those designed from an R&D perspective, because of their use in regulatory matters. In practice, that means, for example, that you can’t switch off the need for signatures.
There are also challenges in using ELNs in protein drug discovery because is it less structured and has more processes than small molecule drug discovery, Dr. Meinhof explained. “In protein drug discovery, we will need more interfaces, each of which will be useful to fewer people,” than would the structured, standardized interfaces used for small molecule discovery. ELN flexibility, including the ability for scientist to develop their own templates, he said, is, therefore, mandatory.
The absence of a standardized way of defining the structure of proteins and the cell lines that make them, is a problem that affects all experiment records with biological entities, according to Dr. Meinhof.
What ELNs Need
“Such diversity of data types requires a variety of specific technologies for managing and searching such data,” Dr. Hemmer said. So, although ELNs do handle different types of data, “all data doesn’t belong in an ELN system.” The ELN exists to create context for raw data. Thus, he advises against systems that try to do it all. “You have to deal with different types of information differently,” he pointed out, noting that “you can’t search for structures as if they were text, and a simple identity search for experimental spectra would be unsuccessful because of small deviations.” One modular platform can succeed.
“Currently, some ELNs focus on single experiment records, treating them as isolated entities,” Dr. Meinhof elaborated. The reality, though, is that scientists collaborate with colleagues and need to easily exchange information in each other’s notebooks. Consequently, Dr. Meinhof said scientists would benefit from having the ability to hyperlink among internal data sources. “This would allow links between two experiments to work bidirectionally (like a reverse-citation index) rather than uni-directionally, like typical Web hyperlinks.
“An analyst who analyzes a sample might want to reference the experimental record that describes how the sample was made. This backward reference would then automatically provide the creator of the sample with a forward reference to its analysis record,” Dr. Meinhof said.
He also sees a need to enhance data-extraction functionality to minimize duplication in ELNs, LIMs, analysis software, reports, and other applications. Amgen plans to “rely as much as possible on the ELNs as the single data-collection application,” Dr. Meinhof said, so ELNs are the data-entry “front-end for databases that store discrete data” as well as the way records of experiments are created.
Contur’s system is designed around the concept of local, rather than centralized, management. This limits dependence on IT departments, which may be fine for individual workgroups, but less satisfactory for the enterprise. “We give our customers the tools to manage on their own, using smart tools to decentralize the ELN and bring it down to a team level,” said Rozlucki. For this to be effective, there is a caveat. “You have to set minimum standards for projects to ensure that the data is accessible throughout the enterprise, and you have to have a way of taking advantage of local databases and department-specific systems.”
Data storage is another consideration, stressed Charlie Sodano, Ph.D., manager of information services, Bayer HealthCare Pharmaceuticals (www.pharma.bayer.com). “Data needs to be kept for 30 years, but software and hardware are changing daily.” By the time the data needs to be accessed, the machines to run the software and access the storage medium probably won’t be available. To put that in perspective, try accessing a 1980s document written in WordPerfect v 1, stored on a 5-inch diskette. “If it’s really important, put it on paper after the experiment is judged to be completed,” Dr. Sodano advised.