Biography

Lemon was born, and still lives, in the South Hampshire area of Southern England. He gained a first class degree in chemistry with computer science from Reading University and a PhD in Computational Chemistry from the University of Bath on 'Modeling the biological membrane'. He took a Post Doc position at Southampton University on 'Side Chain Placement algorithms' and then joined chemistry software publisher MDL in 1996 where he remained until the end of 2000, working on combinatorial chemistry and Markush search systems.

He was asked to join IDBS as head of chemistry development at the end of 2000 and built a chemistry development team delivering chemistry applications and statistical modeling tools to leverage integration to the biological data in IDBS other products.

In 2005, he and co-founders Ted Hawkins, Alec Gibson, and Robert Shell formed The Edge Software Consultancy, a multidisciplinary consulting company helping pharmaceutical and biotechnology companies to meet the challenges of modern drug discovery informatics. Their knowledge covers biology, chemistry, data management, and computational science, and in the company's first year they signed partnerships with Teranode, Inc., and Chemaxon to deliver new solutions to the industry, alongside developing their own consulting services products. Today, The Edge works with several global pharmaceutical companies helping them increase the efficiency and productivity of their software tools.

Questions

Is the Edge, as in "leading edge" or something else?

"The Edge" is designed to convey our approach to discovery informatics challenging traditional solutions and raising the bar on what is expected from organizations such as ours.

What drives The Edge?

We are all creative people with lots of experience; bringing this to bear to improve the lives of scientists and solving some of their issues with IT tools is what drives us.

Why do scientists need management of experimentation?

The complexity and diversity of experimentation during drug discovery and development only increases with time. Many areas are still not served with good software tools that help scientists to meet the challenges of changing dimensionality and uncertainty during the course of an experiment. Managing all this data, and making it accessible and linked is the key to approaches such as translational medicine. Learning from every piece of data collected and then applying that knowledge across the whole process is going to be the key challenge.

Aren't electronic laboratory notebooks (ELNs) analogous to the paperless office? A nice idea but something that's not going to truly happen across the industry?

Tell that to the hundreds of pharmaceutical and biotechnology companies facing the increasingly difficult task of managing intellectual property. Paper notebooks are great for recording information but the end result is a dead document that isn't searchable and the provenance of the data is essentially lost without extensive manual efforts. Increasingly, companies are realizing that valuable data is simply lost when a scientist moves on or retires. There is a strong belief that ELN can help to capture this information. The reality is, as always, somewhere in between, some information can be captured but the real connectivity and knowledge will always be in the scientist's head. I'm a firm believer that if you take the serendipity out of the process, the result will be fewer new molecular entities (NMEs). This reduces the Darwinian element of the process.

So, what are the real needs for ELNs?

It's simple really, scientists need tools that match the way they work, not tools that expect them to change the way they work to match the software. Meeting this challenge is going to be the single largest factor in the success of the ELN movement. We have developed a method of understanding and documenting the key requirements for usability and productivity and delivering it in a way that internal IT teams and developers can deliver that usability. If we can get this right, the days of the paper notebook are numbered, in 10 years from now we'll laugh that we used to use paper to record scientific observations.

Tell me about Markush structure handling.

Many people use the term Markush to refer to combinatorial libraries which are collections of real molecules described in a compact form. A true Markush structure is far more than that. A Markush structure is designed to both encode a real collection of molecules and also cover a more abstract area of chemical space that defines the intellectual property of the patent. It's one of the last peaks to be conquered in informatics. Much work has been done on this in the past and I'm seeing a renewed interest in the subject recently from companies such as Chemaxon.

Currently the industry's ability to search and mine the patent data is limited to a few specialist people who are capable of using the existing systems, many of which date back 15-20 years. What's needed is a modern solution to the problem where scientists can search and mine the patent data themselves and open up the wealth of knowledge that's available to make use of. Imagine being able to search the patent literature during novelty checking when designing a new molecule, or being able to examine potential side effects from similarity searches in known patents against particular diseases. This is far from a trivial problem, I have some ideas in this area which I'd like to pursue again one day...

In what ways can software provide a total solution to the burgeoning biological data with which scientists are faced on a daily basis?

The present situation is very similar in many ways to other industries experiencing increased automation and scale. Well designed software can help to automate routine tasks and increase productivity. It can take away the mundane and enable clever science. However, all too often these days software is designed by software engineers for people like themselves and dealing with the software becomes as bad if not worse than the original mundane task. We concentrate on helping organizations buy or build solutions that meet the scientist's needs and are not just technology led.

In what ways can business analysis in the life sciences improve work and information flow?

Any IT project lives and dies on its requirements. A mistake in requirements can be 100 times more expensive than a bug in the software. We have developed a set of techniques for capturing requirements and backing them with evidence from the scientists themselves. This takes the emotion and guesswork out of selecting the right tool or implementing the solution.

What's your method "for analyzing and capturing an understanding of the real needs of scientists"?

We have developed a methodology based on a number of established techniques which involves interviewing scientists and capturing not just what they do, but their goals and aspirations. We aim to capture the essence of the archetypal users. We give them a name and characteristics. Personalizing the mythical user helps everyone in the project to focus on what the solution needs to deliver and, more importantly, why. From this we develop use cases and put them into the context of the typical users. From this information requirements for the system have sufficient life and depth to remove some of the risk of traditional software approaches.

In what ways are your experiences with MDL and IDBS allowing you to build Edge?

As a fresh-faced computational chemist I learned about cheminformatics at MDL, it was a great place to learn and understand how to build software and some of the challenges of dealing with chemical information. IDBS was an excellent education in business. I learned to run my product line like my own company. I've been lucky enough to work with some great people and have learned that in a good team "the sum of the whole is worth more than the sum of the individual parts!" So when we formed The Edge it was a coming together of some great people and time to put in practice all we have learned from our collective experience. So far we have found our enthusiastic and professional approach is appreciated by our clients and we're able to exceed their delivery expectations.

Are they critical to the tools and services you're now offering or are you in a position to work with other companies and to what end?

A good question, this is one of the traps many consultants can fall into. Our approach builds from our knowledge of the industry not from any particular software. I realized very early on in my career that to base ones' skills around any particular system is a big mistake, better to understand the nature of the problem being solved which is reusable knowledge. We offer solutions and services all aimed at saving one day per week per scientist and to that end we'll work with any software or indeed create our own to achieve that goal.

Have you had any big successes you could tell me about?

I'd love to tell you all about the project we're working on at the moment, but as a consulting company we have to respect our client's confidentiality.

What obstacles do you foresee in convincing the industry of the value of your approach?

In this industry experience is everything. As a consultancy company you live and die by your last assignment. We strive to exceed our clients' expectations every time and so far we have a 100% repeat business rate which speaks volumes for the value of our services. Another is innovation. Surprisingly, there is actually not that much innovation going on in the established commercial companies in my view. As an independent consultancy, we're able to survey and keep up to date with many new trends. We're working with Teranode to explore application of Semantic Web technology within the life science domain which is very exciting.

Other than work, what should scientists do on that "one day a week" you promise to reclaim for them?

Walk to a tall mountain and scream your lungs out, it's not good for global warming but its great for the soul.