The above thought was inspired by the following problem that an Accelrys field scientist presented to me: Given 50 animal subjects with varying body mass, how can they be divided into 5 equal-sized groups such that the body mass mean and variance are roughly the same for each group? In other words, we want the distribution of body mass within each group to be nearly the same. To both of us, this looked like a Pareto problem in which the variance of the within-group mean (variance-of-mean) and the variance of the within-group variance (variance-of-variance) across groups should be simultaneously minimized.

I implemented a simple genetic algorithm in a Pipeline Pilot protocol to do a Pareto optimization for this problem. Here’s what the typical results look like in a tradeoff plot showing the Pareto results after 500 iterations, compared to results for randomization and minimization (both constrained to result in equal group sizes):

Variance-of-variance vs. variance-of-mean for alternative methods of assigning subjects to groups

Each color/symbol represents a different approach, with results from Pareto optimization shown as blue circles; results from multiple random partitionings shown as black stars; and results from minimizations with the subjects processed in different random orders shown as red triangles. Observe that for this very simple problem, the Pareto approach gives the lowest variance-of-mean and variance-of-variance values.

Might Pareto optimization be an alternative to traditional randomization and minimization for trial design? I don’t know. Given the constraints of the approach, it may be more applicable to lab studies than to clinical trials. But these exploratory results look intriguing. I provide the protocol along with some more thoughts in an Accelrys Community forum posting.

The new functionality in this release is focused specifically on process chemists, who work in areas that are strictly regulated and subject to stringent quality assurance (QA) oversight. The greatest challenge faced by process chemists is identifying, optimizing, and delivering the best process to pilot and manufacturing plants in the shortest time possible while complying effectively and efficiently with regulations and QA mandates.

As described in the press release, the 6.5 version of Symyx Notebook by Accelrys offers an array of new features requested by customers already working in regulated environments. Crucially, the system offers the ability to search experimental data stored in most legacy Symyx notebooks. Such functionality is clearly valuable to those who have these legacy Symyx notebooks. But the underlying API also provides an opportunity to develop a single portal to any ELN data—whether that ELN is from Symyx or another ELN vendor.

Visit www.symyx.com/elnprocess to learn more about how the 6.5 release of Symyx Notebook by Accelrys supports process chemists. Representatives from AstraZeneca and Bristol-Myers Squibb discuss their use of an ELN in process chemistry in a downloadable PDF and a short video recorded at the Symyx Symposium. You’ll also find an informative slideshow describing the release. We also encourage you to check out our sibling blog, Symyx’s Life in the Electronic Lab, which will feature several posts over the coming weeks on ELN use in process chemistry and the 6.5 version of Symyx Notebook by Accelrys.

Over the past year several of my fellow scientists and I have been investigating the ways that both modeling and informatics solutions can be applied to CPG R&D. There are some great write-ups out there:

• Speeding Flavor and Fragrance Innovationby Drs. Gestoso-Souto and Subramanian
• Modeling can Improve Your Products’ Look by Dr. George Fitzgerald
• Transferring Innovation to Global Markets and Smart Innovation Reaches for Relevance by Dr. Michael Doyle

Also note Dr. Felix Grant’s article Material Values in Scientific Computing World. And check out the great image on the cover.

What’s fueling this recent surge of scientific informatics and modeling into CPG? Like so many other R&D-based sectors, CPG has been challenged to remain competitive while holding down costs — and nothing says “take my money” like a wrinkle cream that really works, and is sold over the counter at the local pharmacy! As the articles I’ve cited illustrate, predictive analytics can help R&D teams find answers faster and for less $$ than can experimentation alone – a lesson that sectors such as Pharmaceuticals learned quite some time ago. Add to this the that predictive (molecular) modeling methods have become easier to use and are increasingly merged with informatics, and you gain unprecedented R&D capabilities.

Modern pharmaceuticals are extensively modeled before chemicals are ever mixed in the lab. Maybe the next great perfume will come out of a computer simulation, too.

A recent review article on this topic by long-time associate Prof. Richard Catlow, et. al, caught my attention. Readers of this blog will be familiar with our many posts pertaining to ‘green chemistry,’ sustainable solutions, and the like. Last month, Dr. Misbah Sarwar of Johnson Matthey was featured in a blog and delivered a webinar on the development of improved fuel cell catalysts. Dr. Michael Doyle has written a series on sustainability. Drs. Subramanian and Goldbeck-Wood have also blogged on these topics, as have I. All of us share a desire to use resources more responsibly and to ensure the long-term viability of our ecosphere. This will require the development of energy sources that are inexpensive, renewable, non-polluting, and CO2 neutral. Prof. Catlow provides an excellent overview on the applications of molecular modeling to R&D in this area. Read the paper for a very comprehensive set of research problems and case studies, but here are a few of the high points.

• Hydrogen production. We hear a lot about the “hydrogen economy,” but where is all this hydrogen going to come from? Catlow’s review discusses the generation of hydrogen from water. Research challenges include developing photocatalysts capable of splitting water using sunlight.
• Hydrogen storage. Once you’ve created the hydrogen, you need to carry it around. Transporting H2 as a compressed gas is risky, so most solutions involve storing it intercalated in a solid material. LiBH4 is a prototypical example of a material that can reversibly store and release H2, but the process is too slow to be practical.
• Light absorption and emission. Solar cells hold particular appeal, because they produce electricity while just sitting there (at least in a place like San Diego; I’m not so sure about Seattle). One still needs to improve conversion efficiency and worry about manufacturing cost, ease of deployment, and stability )with respect to weathering, defects, aging, and so forth).
• Energy storage and conversion. Fuel cells and batteries provide mobile electrical power for items as small as hand-held devices or as large as automobiles. Catlow and co-workers discussed solid oxide fuel cells (SOFC) in their paper. 

 The basic idea with modeling, remember, is that we can test a lot of materials for less cost and in less time than with experiment alone. Modeling can help you find materials with the optimal band gaps for capture generation of photoelectric energy. It can tell us the thermodynamic stability of these new materials: can we actually make them and will they stick around before decomposing.

Simulation might not hit a home run every time, but if you can screen out, say, 70% of the bad leads, you’ve saved a lot of time and money. And if you’re interested in saving the planet, isn’t it great if you can do it using less resources?

Check out some of my favorite resources on alternative energy, green chemistry, and climate change.

• ACS Cleantech Chemistry
• Advancing Green Chemistry
• Alternative Energy News
• Skeptical Science

A big challenge for the fragrance and flavors industry, not unlike many others, is that regulatory changes continue to reshape how business is done. As pricing pressures on raw materials continue to drive up prices companies need every possible means to drive smarter, more efficient innovation.  In the midst of this, these companies must focus on those areas that their clients need to compete and grow – one such way is through more creative, differentiating flavors and fragrances. 

Globalization and emerging markets such as China, India, Latin America, continue to provide new opportunities for growth.   However, to succeed in these emerging markets one cannot just reuse successful existing flavors and fragrances since the consumers will have a different aesthetic taste.  Continuous innovation is needed to be successful in these global markets.  It is a big mistake to think that every chemical and formulation has already been invented.  Smarter, faster and more rational design of new compounds is necessary—and possible—for thriving and growing.  Following in the footsteps of the life science industry, who have been using predictive techniques for years, more and more consumer products companies are turning to predictive science to achieve their innovation goals and stay in the game (click here to see my article in Perfumer and Flavorist).

The central question Russell raises is “how far workflow tools can grow beyond personal productivity instruments into what Frost & Sullivan has termed scientific business intelligence platforms serving the enterprise.” I agree this question is key, but it is sometimes misinterpreted. “Serving the enterprise” is often equated with moving from providing software for an individual scientist to an ill-defined concept of an “enterprise user” and some fuzzy concept of increased organizational performance. We don’t necessarily see it that way.

Let’s say I’m a scientist at a pharma or biotech company, and I create a Pipeline Pilot data analysis protocol to automate and standardize what I previously did manually. That benefits me, but I can also share that protocol with my colleague sitting next to me, and she gains a similar benefit. Well, why should seating proximity determine who gains from my standardized process? In fact I can share my protocol widely with colleagues throughout my organization, across geographies and time-zones. So now we see clearly that the organization benefits both in terms of collective enhanced personal productivity, but also globally, due to the standardized application of analytical methods leading to more consistent data that ultimately supports better decision-making across projects and over time.

This is a key element of what we consider “scientific business intelligence.” More and more, the ability to transfer expertise across the enterprise has become a requirement. Data pipelining and workflow naturally facilitate master data management and good practices for data stewardship that benefit organizations internally and in their necessary collaborations with CROs and academics. These systems help enable organizations define consistent methodologies that can be extended and enforced both across and between organizations to ensure that data sets are comparable regardless of who creates them. Pipeline Pilot also scales while conforming to IT policies around security and authorization, which supports the growing demands for personal productivity on an enterprise scale. It’s this “enterprise readiness” that has made Pipeline Pilot the pipelining tool of choice at 20 of the top pharma companies.

Russell also notes the apparent hype associated with data pipelining/workflow tools. I’m not sure we really see or hear an excessive amount of “hype,” but some of the quotes in the article are telling. “The best possible data management tools would provide 10% of the solution, at most.” Whether this is the right value or not, it is a fact that significant inefficiencies remain in our industry due to the continued use of Excel copy-and-paste and similar manual, repetitive, error-prone processes. As another quote states, “Eventually, workflow tools can and should supplant many data analysis processes that are currently done with a combination of fragile spreadsheets and marginally re-usable macros.” This is consistent with estimates that more than half of a researcher’s time is spent performing these manual, data processing tasks—tasks that are highly amenable to being coded into workflows such as Pipeline Pilot protocols (e.g., R&D Productivity Analysis, Yazdani, Holmes, A.D. Little).

 So we shouldn’t be thinking in terms of how much of the business of drug discovery (or other science and technology-based research) can be provided by workflow tools, but rather the extent to which these tools can minimize the time spent by scientists performing routine tasks and free them up to actually innovate and be productive within their organizations. This is, I believe, one of the “towering aspirations” of workflow tools. If we could reduce the time spent in low-productivity tasks to an amount closer to 10%, it would provide a massive benefit to research organizations. Wouldn’t this make workflow tools worthy of just a bit of hype?

“In recent years there has been a lot of interest in fuel cells as a ‘green’ power source in the future, particularly for use in cars, which could revolutionize the way we travel. A (Proton Exchange Membrane) fuel cell uses hydrogen as a fuel source and oxygen (from air), which react to produce water and electricity. However, we are still some time away from driving fuel cell cars, as there are many issues that need to be overcome for this technology to become commercially viable. These include improving the stability and reactivity of the catalyst as well as lowering their cost, which can potentially be achieved by alloying, but identifying the correct combinations and ratios of metals is key. This is a huge task as there are potentially thousands of different combinations and one where modeling can play a crucial role.

As part of the iCatDesign project, a three-year collaboration with Accelrys and CMR Fuel Cells funded by the UK Technology Strategy Board, we screened hundreds of metal combinations using plane wave CASTEP calculations.

In terms of stability, understanding the surface composition in the fuel cell environment is key. Predicting activity usually involves calculating barriers to each of the steps in the reaction, which is extremely time consuming and not really suited to a screening approach. Could we avoid these calculations and predict the activity of the catalyst based on adsorption energies or some fundamental surface property? Of course these predictions would have to be validated and alongside the modeling work, an experimental team at JM worked on synthesizing, characterizing and testing the catalysts for stability and activity.

The prospect of setting up the hundreds of calculations, monitoring these and then analyzing the results seemed to us to be quite daunting and it was clear that some automation was required to both set up the calculations and process the results quickly. Using Pipeline Pilot technology (now part of Materials Studio Collection) protocols were developed which processed the calculations and statistical analysis tools developed to establish correlations between materials composition, stability and reactivity. The results are available to all partners through a customized web-interface. 

The protocols have been invaluable as data can be processed at the click of a button and customized charts produced in seconds. The timesaving is immense, saving days of endless copying, pasting and manipulating data in spreadsheets, not to mention minimizing human error, leaving us to do the more interesting task of thinking about the science behind the results. I look forward to sharing these results and describing the tools used to obtain them in more detail in the webinar, Fuel Cell Catalyst Discovery with the Materials Studio Collection, on 21^st July.”

Of course, as computers get faster, the sorts of calculations that customers want to run also get larger and more “realistic.” The explosive growth of multi-core architectures has lead to an increased focus on parallelization strategies to make the most efficient use of the CPU power. Traditionally, the main focus with high performance codes has been on getting them to run in fine-grained parallel as efficiently as possible, scaling a single job over 100’s of CPU’s. The issue here is that, if you are only studying a medium sized system, you will quickly get to a point where communication between nodes costs far more than the time in the CPU, leading to poor performance.

The solution, if you are doing multiple calculations, is to run multiple calculations in a coarse-grained, or embarrassingly parallel, approach. In this mode, you run an individual calculation on a single CPU but occupy all your CPU’s with multiple calculations. Materials Studio will do this but it can be monotonous, although you can use queuing systems to get somewhere here. The other issue is that when you start doing many calculations, you also get lots of data back to analyze. Now you not only need a way to submit and control all the jobs, but also to automatically get the results you want.

Luckily, the Materials Studio Collection in Pipeline Pilot has both of these functionalities. You can use the internal architecture to submit jobs in both fine and coarse grained parallel (or a mixture of the two if you have a nice cluster!). You can also use the reporting tools to extract the data you want and display it in a report.

A great example of this is working with classical simulations of polymers for calculation of solubility parameters. Here, you need to build several different cells to sample phase space and then run identical calculations on each cell. You can quickly develop the workflow and then use the coarse-grained parallel options to submit a calculation to each CPU. When the calculations are finished, a report of the solubility parameter can be easily generated automatically.

So, whilst there is a definite need for an expert modeling client like Materials Studio, the Materials Studio Collection really does free you up to focus on problem solving (using Materials Studio!), not waiting for calculations to finish and then having to monotonously create reports!

The full press release explains more about the merger, as does this page on our website. The team leading the transition is committed to supporting the investment you have made in products from both of our companies. Max Carnecchia will continue on as Accelrys CEO and Isy Goldwasser, CEO of Symyx, will continue as an advisor through the transition. The balance of our management team includes members from both companies. Customers should continue to request help from the respective Symyx and Accelrys technical support teams via existing phone, email or website portals.

We know you have questions. In addition to the links above, we have created a new area in the Accelrys Online Community called The Headquarters. This group is open to customers and partners of Accelrys and Symyx, and we encourage you to join the discussion with our team members.

We look forward to hearing your thoughts on this exciting development!

The paper sparked a series of debates about the significance of this particular experimental strategy and to what extent it constituted the creation of synthetic life. Additionally, and of more practical interest, the efficiency of this de novo strategy was also widely debated. While there is relatively little dispute about the potential value of engineering new biological systems, the work fuelled an ongoing controversy about the efficiency of such an empirical approach to creating a new biological compared to more conventional strategies that modify existing genomes in their native cellular environments.

But the study also emphasized some more specific challenges beginning to emerge as genome sequencing establishes itself as a mainstream tool in the Life Sciences. Of the many technical obstacles that the team overcame in creating the new cell, one of the most surprising was the impact of seemingly trivial error rates on the successful creation of the synthetic genome sequence. A single base pair deletion in the dnaA gene, involved in chromosome replication, rendered the transplanted cells unviable. The failure to identify this error during quality control sequencing of the synthetic genome significantly delayed the completion of the project.  As soon as this one-in-a-million error was identified and rectified in the synthetic sequence, the team was able to successfully recover viable cells.

In this case, the significance of the sequencing error in determining the synthesized genome sequence was obvious, as it resulted in an effectively lethal genotype. However, more generally, it emphasizes the critical need for highly accurate sequence determination. As the pharmaceutical industry increasingly relies on genome sequence data as a foundation of personalized medicine, the accuracy of the genotypic data collected on a large scale will come sharply to focus. Being able to determine sequences with 100% accuracy, particularly in non-coding regions of genomes, may become a challenging pre-requisite to personalized therapeutic strategies.

Accelrys is actively working on products, such as the Next Generation Sequencing (NGS) component collection and a biological registration system that can play a critical role in quality control of biological data. In the case of the NGS collection, alongside the integration of the latest mapping and assembly algorithms, the core data pipelining capabilities of the Pipeline Pilot platform make it possible for scientists to develop quality control pipelines without any programming knowledge. As applications of genome sequencing such as synthetic biology and personalized medicine progress, such computational approaches to quality control will play an increasingly central role.