Sunday, May 31, 2009

Streamlining Scientific Processes

Get out of the silos

In the last decade, research in drug discovery and the life sciences has been dominated by two trends: increased automation and greater parallelization. Once, a researcher would focus on cloning a single gene at a time. Now, with the advent of robotic technology, they can clone thousands of them in parallel. Similarly, researchers no longer attempt to crystallize a single protein, instead screening dozens of variants against hundreds of conditions simultaneously. High-throughput screening allows us to screen millions of potential drug leads against hundreds of targets, and single nucleotide polymorphism (SNP) genotyping enables the search for millions of genomic variants. In discipline after discipline, automation has fattened our pipelines by allowing us to simultaneously study many items in parallel.

With this increased capacity comes new organizational and IT challenges. On one hand, individual departments are overwhelmed by terabytes of data. In order to function effectively, they need immediate solutions to track the huge number of items running through their pipelines. On the other hand, management faces new challenges to integrate the data from multiple departments. They need rapid insight into project status and resource requirements for these massively parallelized operations. Management must also be able to quickly identify bottlenecks, so they can plan effectively and commit wisely.

Too often, individual departments take it upon themselves to address their specific needs and problems using independent IT solutions without considering the effects on the overall enterprise. And why shouldn’t they? Their needs are immediate and essential because if they don’t manage to accurately track the voluminous data, all the work is for naught. The drug discovery staff need to track protein data, so they build Excel spreadsheets to solve their immediate problem. The genotyping people need to track clones, so they set up an Access database. This independent approach gets replicated for different problems across different departments. But despite the fact that these piecemeal solutions all describe common and shared items, no one puts those departmental solutions together in a cohesive workflow. Instead, each solution remains separate and inaccessible. Information remains trapped in isolated environments, instead of flowing between departments, applications and users.

This piecemeal, silo’d approach, where departments coordinate by transferring separate tidbits of information, no longer works in a high-throughput, parallel environment. Instead, true success is now measured by how well critical data is captured, communicated, and tracked within the organization. That critical flow needs to start at basic research and move smoothly through clinical development to manufacturing in a timely, secure and efficient manner.

When management needs to know what is going on with a particular project, the information has to be readily available. At one small company where I worked, keeping track of all the supporting data to determine the status of a client’s crystallography project was a nightmare. Each quarter, we had a fire drill to quickly assemble all the information we needed to determine which samples worked and those that did not. Without an enterprise-wide system to track the data, we needed to email our colleagues in a back and forth manner — a process that was simply inefficient and a headache for everyone. We understood how to do basic sample tracking, but not comprehensive project tracking.

What we learned from that experience was that we needed a new approach, one that links applications and data across multiple functions to streamline and automate workflows and processes across the organization.

The solution was integrated enterprise software that enabled research organizations to fully leverage their knowledge assets. More effective collaboration with colleagues can be streamlined with project management and knowledge management modules. Replacing paper-based laboratory notebooks with electronic lab notebook software can automate the process of capturing and organizing information and prevent work redundancy. The result is greater visibility into past knowledge and the ability to see what has worked in the past and what hasn’t — all with a minimum of finger-pointing.

These strategic changes require getting everyone at various levels involved so they know and understand what is going on in the process. Having access to the information makes people accountable. Tasks are more likely to be completed on time. Communication is greatly improved with less overhead and lower frustration. Most importantly, giving everyone access to information leads to the right and best decisions.

Not only does this 21st century bioinformatics streamlining provide improved workflow for completing projects with greater accountability, it also enables everyone with access to have confidence in the scientific data. When building organizational-wide information technology solutions, there are several techniques and issues to consider:
  1. Start with meta-data-based, configurable solutions that provide a flexible data model that stores results for a variety of scientific applications. The best systems don’t require a programmer to do most or all of the configurations.
  2. Pick a web-based system that allows scientific personnel easy access from anywhere within the enterprise, rather than a client-based application.
  3. Implement solutions that have an open API or service-oriented architecture (SOA) for easy integration with other systems such as labware and LIMS (Laboratory Information Management Systems).
  4. Use an integration strategy in which the software manages the information tracking and data retrieval. Make sure the software handles cross-departmental communications – a key factor as more biology lab processes are becoming automated.

Of course, pursuing this enterprise-approach to workflow computing requires dealing with company-wide change. Often organizational dynamics are the biggest barrier to acceptance and adoption. It requires a good clear vision from scientific leadership that an information integration strategy is necessary for the company to operate efficiently. Without this management directive and support, it can be difficult to overcome organizational inertia. Make no mistake; there will be roadblocks, because the bigger the project, the scarier it gets – especially for people who sign up to make it happen.

The best advice: Carefully choose a platform that will allow an enterprise-wide integration up front, but within that solution show you can solve one department’s needs effectively and thus enable people to become familiar with this new approach so they can feel secure and successful. As you move on to other departments, you will be planning to solve company and division issues while also handling departmental needs. By starting small at one level, and choosing a platform that you can grow with, you can achieve your goals of integrating your systems in the end.

No comments: