Leveraging research informatics to accelerate drug discovery

By R Arun Kumar

The main drivers of research innovation in the Life Science industry are patent expiries, pricing pressures, evolving therapeutic needs and the advent of biologics for drug development. The need to find new drugs in the fastest way possible has become more urgent. Geographic spread of research endeavors, new research methods, overflow of scientific research data, and the need for collaborative research practices are defining today’s research environment. In spite of having at their disposal advanced research techniques such as Genomics, Proteomics, Marker Based Assessment, and Microarray Technology, organizations find it challenging to realize optimal research outcomes. This paper presents various research challenges and ways to maximize the value of Research Informatics investments.

In recent years the pharmaceutical industry has declined in performance (1), with replenishment of the product pipeline becoming the main criterion for drug discovery research transformation. With US$ 60bn worth of products going off patent by 2011, life science companies must identify novel and innovative methods to compensate for falling research productivity (2). According to Jean-Pierre Garnier, CEO GSK, “broad transformations of the organization are necessary first steps…only the very best players will be able to meet the challenge and rebuild their R&D engines” (3).

The emerging challenges faced by researchers have outpaced efforts to address them. Research labs are generating data faster than can be integrated. Life science companies are adopting omics-based scientific methods to gain information and knowledge on target validation. While the pharmaceutical industry has been adept at optimizing the drug development process, it has rarely implemented different structures (4) to make the discovery process more efficient (5). Therefore, research industry and its technology service providers must constantly innovate to support the needs of emerging science.
Trends in discovery research processes
• Adopting biological discovery methods or marker-based assessment to unravel new therapeutic solutions – build emerging core processes.
• Connecting molecular biology results with clinical research outcomes to conduct root cause analysis of a disease and its response to therapy – feedback loop processes from distinct disciplines.
• Instituting standard methods and activity procedures in most of the innovation-led project operations – follow standardized workflows.
• Using cross-disciplinary scientific research processes for drug discovery – reduce ambiguity in terms and terminologies within chemistry or biology.
• Collaborating with external partners and even competitors for scientific discovery to solve a common research problem.

Challenges in discovery research

Scientific research generates data by registering biological or chemical entities and testing their biological, physical or chemical character or their pharmacological action. Information related to registration and assay workflows forms the basis of all scientific innovation in any disease program. However, there are several challenges including process complexity, data indecipherability and questionable technology efficacy.

The following table categorizes the most critical issues:

Fig. 1: Broad classification of challenges in discovery research

Multiple workflows

Multinational pharmaceutical companies have globally distributed laboratories engaged in related or similar research activities. Mergers and acquisitions lead to redundancy and lack of harmony in laboratory workflows. Laboratories belonging to multitudinous disciplines of biology and chemistry practice variations of the assays or registration workflows. Some of their activities are centered on hypotheses requiring a composite set of processes to be orchestrated together. Many of these laboratory processes are dependent on the instruments used for analysis. Multiple workflows hinder distribution, reuse and adoption of best practices within the global scientific community.

Research work in silos

Systems biology and chemical screening centers of excellence supporting mainstream programs are located in places with easy access to infrastructure and a low-cost talent pool. Traditional chemistry and emerging biology research must be integrated to create an inter-disciplinary activity. A lack of collaboration between biology and chemistry processes leads to repeat and redundant work and inconsistent results. Integrating chemistry data within the context of biochemical processes or integrating genetic data with biological pathway information can facilitate better understanding of disease. Hence there is a need for biology and chemistry experiments to “cross-over” in the interest of disease research. A chemistry-biology interface would be required in structural biology, enzyme function, structure function, or protein folding research in order to state the disease problem better.

Large volume of information

Research laboratories generate large volume of scientific data to draw insights and inferences. Recent experimental techniques such as the omics methods and computational simulation generate terabytes of raw data in its every run. Effective analysis and annotation of the basic reads or data sets can be laborious without the help of parsers and graphical and visualization tools. Additionally, there are concerns about the security of data stored and exchanged across laboratories and the possible theft of intellectual property.

Heterogeneous data formats

The primary entities in drug discovery research are Diseases, Pathways, Proteins (along with their interactions) and Genes. These are the foundation stones on which new molecule research is built. The biggest obstacle to the integration of research information is that data is usually available in heterogeneous formats and stored in silos, and hence cannot be shared easily. In addition, frequent duplication of information or ambiguity in terms adds to the difficulty of making timely informed decisions.

Fig.2: Research challenges during the drug discovery process

Ways to overcome research challenges

The use of techniques such as data semantics, visual analytics, collaboration and workflow streamlining in drug research can increase research effectiveness, improve predictability, foster team work among scientists and thereby produce better research outcomes.

Streamline process workflows

Pharmaceutical companies continuously optimize their processes and workflows in existing fields of research and adopt new ones to drive innovation in discovery. Procuring new products before studying their alignment with research processes only adds to license fees and maintenance costs, without adding value to knowledge capabilities. One way to optimize costs is to establish a tight linkage between processes and applications. The objective is to select processes that will improve competitiveness, prioritize business activities and enable IT solutions. Pharmaceutical companies should harmonize research workflows, reduce redundant processes and leverage an optimal composition of applications to process information.
• Put together regular biology and chemistry methods as standard routines
• Upload standard workflows onto an electronic system for all laboratories of an organization to follow
• Define workflow boundaries that can help join different disciplines of science

Collaborative research

The therapeutic product portfolio is being increasingly crowded by large rather than small molecule-based screening and optimization. Therefore, erstwhile chemistry and new biology research work closely together in inter-disciplinary projects. Research scientists from different disciplines need to actively understand and address various facets of the disease problem together. An example of collaboration is when results for cell line-based screening assays against a class of inhibitor compounds are jointly interpreted by a biologist and a pharmacologist.

Currently, there is only a moderate level of collaboration. Scientists are not known to freely share and exchange concepts or findings from their experiments or computations. Most often, interchange of ideas and information sharing happens via handwritten notes, whiteboard or electronic mail. For research collaboration to yield meaningful benefits, it must be viewed as an imperative.
• Make collaboration an attractive proposition to scientists, and include it in their key performance indicators
• Technology-enable collaboration by leveraging RSS, Wiki and Online Portals
• Consider using a multi-site collaboration platform that will streamline document workflow, and offline/online content sharing in standard templates
• Promote open collaboration with academia and establish pre-competitive collaboration with industry while safeguarding intellectual property

Visual analytics for large data sets

Scientists evaluate a hypothesis by gathering large volumes of multi-dimensional data for inspection. While raw alphanumeric data can be cumbersome to handle, a pictorial rendition can facilitate analysis. Even as scientists slice and dice through mountains of 2-D graphical data, they often need other types of graphical presentation for drill down analysis. Sometimes, two sets of data are compared and contrasted keeping one of two parameters constant.
• Scientists should be equipped with visual analytics tools that include pre-defined protocols for contextual data mining and filtering
• Choose technologies that support vector-based rendering of millions of data points to show depth, perspective, and performance of 3D image, network, molecule, or data distribution
• Visually represent semantically joined concepts over a life-sized surface display

Semantics for data interoperability

Clearly, aggregation of information across the discovery value chain is pivotal to creating an integrated discovery engine, a holy grail for leading pharmaceutical companies worldwide. Still more important is the ability to carry out cross-functional search. Effective integration infrastructure enhances the ability to carry out cross-functional search on biological and chemical information categories, crashing time-to-market for new drugs. Re-wiring existing data assets in the context of domain ontology is an effective way to achieve inter-operability. Using semantic technologies, researchers and program directors can discover relationships that enable them to make better and faster decisions about disease targets and drug compounds. Data inter-operability with ontologically linked data sets reduces the time needed to assimilate research findings.
• To enable semantic search, structured tabular data needs to be converted into relationship models as per a commonly agreed upon domain ontology
• Web Ontology Language and Resource Description Framework are technologies that unify and integrate data into machine-interpretable concepts

Fig.3: A schematic representation of a research informatics system Semantic

Building a Research Informatics Ecosystem
Companies looking to overhaul their research capabilities must recognize that this is a major transformation effort. As outlined in the earlier section, there are complex inter-dependencies between data and visualization as well as collaboration and workflow. Besides these, the focus areas critical to realization of transformational value are:
•    Business Value Creation
•    Adoption and accountability within the scientist community
•    Strategic partnerships with external parties
•    Management of the overall program

Conclusion

In the face of major challenges and pressure to replenish the revenue pipeline, research organizations have an opportunity to re-emerge as the growth engines of industry. This will need significant commitment from the leadership and a strong vision for the future supported by the ability to acquire and deliver value in a phased manner. Pharmaceutical organizations, patients, payers and governments alike will welcome faster and cost-effective discovery of innovative therapies for present day medical challenges.

References

1. Market watch: Pharma industry strategic performance, 2008–2013E. Michael Goodman. Nature Reviews Drug Discovery, 8, 348, May 2009.
2. Why has R&D productivity declined in the pharmaceutical industry? R R Ruffalo, Expert Opin Drug Discovery, 1, 99-102, 2006.
3. Rebuilding the R&D Engine in Big Pharma. Jean-Pierre Garnier. Harvard Business Review, May 2008.
4. Optimizing the discovery organization for innovation. Frank Sams-Dodd. Drug Discovery Today, 10(15), August 2005.
5. R&D Efficiency, Tuft Center for the Study of Drug Development. Tufts Univ., Outlook 2009.

Biography

R Arun Kumar is an Associate Vice-President, who heads the Global Life Sciences practice at Infosys and is responsible for the growth and expansion in the Life Sciences domain. Arun has more than 16 years of professional experience in the areas of business-technology alignment, IT and BPO services, global sourcing, strategy & marketing, software product development, wireless and consumer goods. His career spans multiple continents and he has worked in leadership roles in established trans-nationals as well as in start-ups. Arun lives in the San Francisco Bay Area and can be reached at R_ArunKumar@infosys.com