Interfacing LIMS with ELN

Today, electronic support for scientific research from the bench to the product is reality. For many years there has been an organic growth of different electronic systems in various areas of industry. The current challenge is to combine these electronic islands to form solid ground for integrated cost efficient systems. In this short article I’d like to show how the analytical research LIM system at Merck KGaA is linked with the ELN of the medicinal chemical research.

Usually LIM systems (LIMS) have their origin in the quality-related area. This means that their strength is typically the recording of batch related issues. LIMS are typically involved in laboratory sample management and keep track of users, instruments and standards. They are also regularly used for invoicing, plate management and work flow automation. Some extend to the realm of workflow management and raw data archiving. In many cases they can collect the data direct from the analytical measurement device.

Most LIMS in the market lack a support of chemical structures.

LIMS aim to provide a complete documentation of the workflow within the analytical laboratory and to make the business cases transparent to management and regulatory institutions. To increase laboratory efficiency they can provide shorter order latency (less time per order), automatic billing and detailed reports (without additional resources).

LIMS tend to be adapted to their environment. There are several mostly incompatible types of LIMS in the market, for example research LIMS (small versus biological molecules), pre-formulation LIMS and production LIMS.

Electronic Laboratory Notebooks (ELN) are designed to replace the paper notebooks of the bench chemists. They make in-house research results better available to researchers and help to avoid reinventing the wheel. They also help to document the results in less time, more accurately. In ELNs the Intellectual Property (IP) of the company is stored. This implies that strict, fine granular access control to these systems is mandatory to avoid an information drain to third parties.

ELN systems deal with graphic formulas, reactions, metadata of reactions, control recipe, apparatus etc. Essentially, ELNs aim to provide complete documentation of the workflow within the research laboratory, with detailed transparency of the results. To increase laboratory efficiency they also offer detailed reports (without further resources), data mining and automated billing.

Many ELN systems in the market have been built on customer request. This means they differ a lot in the scope of services they can offer.

To summarise, the common goals of LIMS and ELN are typically a complete documentation of the Workflow within the analytical- and research laboratory and an efficiency increase in terms of less time per analytical order, automated billing and the generation of detailed reports (without additional resources). Divergent goals from LIMS and ELN are typically the lack of fine granular transparency of results and the lack of support for data mining on the LIMS side.

For regulatory reasons research data must be kept for 65 years. This means that information should only be stored in data formats that are suitable for archiving. (Not .doc, .xls, but .odt, .jdx, .xml, .txt, .csv, .pdf, .tiff, or .svg) Long term storage should be provided such as TSM, SAM-Fs, or Networker, to keep these data over the lifetime of the technology employed to retain it. Similarly, for regulatory reasons production LIMS data must be kept for 10-12 years. This means that, for economical reasons, formats such as .doc and .xls can be tolerated if it is likely that these formats can be read at that time. Also, long term stable, open formats such as .odt, .jdx, .xml, .txt, .csv, .pdf and .tiff are acceptable for file storage. Media-independent long term storage should also be provided.

Even without these regulatory demands, the raw data from LIMS is such valuable information to researchers, that these data should be held for a long time.

In a typical ELN workflow two or more starting materials are used to form a product. As accompanying data there are items such as presumed molecular structure, reaction recipe, solvent used, Sample Name (ELN-order-No.), customer name, cost centre id, security flags, customer telephone, customer E-mail, urgency flags and others.

Some of these data are valuable information to the analytical staff; others are nice to have; some are irrelevant and some information may be missing and should be provided from other sources. For example, the ERP-system can be a useful source of valid cost centre ids and a corporate identity management system could provide a central authentication service.

A typical LIMS workflow starts with the order being electronically submitted to it. Then the corresponding sample is checked in and the order is processed. The essential difference in orders between research LIMS and production LIMS is that the proposed molecular structure is part of the order and a validated molecular structure is part of the result report. Another huge difference is buried in the workflow. In production LIMS there are well documented and rehearsed routine analytical procedures, whereas with research samples analysis must be prepared for the things to come.

Ab initio there are no standards available for the new chemical entities, against which they can be checked. The plausibility of the proposed structure must be validated until a certain level of confidence is reached. In order to limit costs it is not useful to employ all methods available to substances in an early stage of their development. The decision about which analytical method should be employed for which sample is primarily based at the site of the researcher who submits the order. Nevertheless, if the researcher submits an order to a laboratory that cannot fulfil his demand or inspire confidence, another method in the same laboratory or in another should be assigned to it.

Thus the route taken by a provided sample through the different laboratories, with different methods, can vary depending on the result of the methods used. To add additional complexity to that workflow, it should be possible to provide several samples to one order in different laboratories in order to speed up the order. To provide full cost control to the researcher, additional methods are only added to the order after calling them back.

In most cases only one sample is provided with an order. To reduce the latency within the analytical laboratories, a laboratory sample is split from the original sample after check-in at the laboratory. The original sample is checked out immediately to the next laboratory. The final laboratory is obliged to keep the original sample flask until the order is worked through completely.

The final laboratory finishing its laboratory order is in the driving seat of crosschecking the results with the other laboratory order results and providing a cross-validated order result (report) to the researcher. The LIMS provides several order types to the customer (e.g. spectrum only, spectrum with interpretation, structure verification (with on method) and structure elucidation), which deliver different levels of information at different cost.

To keep the researchers informed of the ongoing order, all raw data are available within the ELN minutes after they have been collected from the analytical instruments. It is possible for a researcher to extend, reduce or stop their ongoing orders from the ELN at the desktop.

The data exchange between ELN and LIMS is performed via database mailbox tables. This approach has proven to be safe and reliable.

The typical data collected per order, per laboratory within the LIMS are: LIMS order data (LIMS-order-no.), split-sample-id, measurement device id, experiment-no., laboratory result report, sample place, billing, raw data, laboratory users, analyst, GMP, status complete order. These data are being held together with the other order data. The laboratory order results are available to view online within the ELN and the final result is transferred to the ELN as it is released from the LIMS.

This means that the researcher knows where the sample connected to the order is, knows which laboratories are actually processing split samples and knows who is working on the sample. Additionally, information regarding the status of processing and – if available – raw data and order results, are displayed on the ELN screen.

One area in laboratory automation that is not to be under-estimated is sample labelling. In any case it should be possible to track the issuer of a sample. For practical reasons a sample should be labelled with a unique barcode, the name of the researcher responsible for the sample and an ELN journal name. To achieve this goal, uniquely numbered (high quality pre-printed) labels from within a certain number-range are used and personalised with the research laboratory manager’s name, room number and phone number. Additionally the researchers write their ELN journal numbers onto the label to identify their samples without using a barcode scanner within their laboratory. This labelling approach has proven very successful compared to printing the label directly from an application, because the main identifier (the barcode) is immune to badly adjusted or dirty thermoprinters within a rough industrial environment, which in turn can create problems with barcode scanning. Also the barcode itself is only used as a unique identifier and serves no other purpose. This helps to keep the barcode number space small and increase the font size, which also helps to ensure the readability of small labels on small flasks, in a laboratory environment.

We also provide selected methods and costs in database tables, which can be chosen by the researcher within the ELN analytical order submission dialogue. Additionally, there are different parameters for certain methods. For example, for NMR the available solvents are submitted, which are maintained by the analytical department that appears within the ELN.

This approach is not (yet) based on Web services or what are also termed ‘Service Orientated Application’ (SOA) interfaces. This comparatively young technique, which provides a ‘middle-tier communication framework’ is still in its infancy. The standardisation of the necessary security framework has just finished and some points are not defined in depth. To make this cost efficient and (still) relatively easy to implement ‘lightweight standard’ up and running, a rock-solid, full-featured, flexible, central corporate identity management is mandatory. This does not come cheap.

If implemented without the security aspect in focus, it might be possible to expose the corporate research IP to ‘google hacks’.

To reduce any risk of IP research drain on the LIMS and at the ELN site, a very fine granular access restriction system is employed which only provides access for users within the application to data which are of relevance to the user’s workflow. For example, if a sample is submitted to only one laboratory, users in the other laboratories cannot see it.

Conclusion

An interface between ELN and analytical research LIMS is a great step towards a fully integrated workplace for researchers. It helps to increase the acceptance of the ELN as an essential tool for their work and reduces the effort of documentation on both sides. The researcher rules the process and by using the benefit of knowing all relevant order data and sample tracking, the average time spend of an order within the analytical department could be reduced significantly (> 20 per cent). The method described is unique as all of those linkages of LIMS and ELN systems are usually unparalleled (even if two companies use the same software it is extremely unlikely that that software is configured in the same way). There is no silver bullet to overcome the complexity which is inherent to those projects, but the effort is certainly worth it. As software tools are emerging and better infrastructure becomes available these linkages will become less expensive and can be used by smaller companies. The deeper integration of the analytical workflow into the research process, and better interface for the researchers, who have comparatively fast interactive access to their results, helps reduce redundant work and speeds up the pharmaceutical development process. Laboratories also benefit from superior transparency of the processes and better documentation than the old paper-based process.