March 3, 2017

Outsourcing of Big Data Analysis

"Many large drug companies have decided that big data informatics are not a core competency, and have elected to outsource this as a service" says David Shaywitz in The Atlantic

Shortly after, in an article on Forbes, the same author says "many large drug companies seem to have decided that hefty big data analytics is a service to be outsourced".

To me, he appears to be confusing two things here. The outsourcing of informatics and analytics are two very separate issues. 

Informatics is an infrastructure, including storage and compute and a range of analytics tools, that can be used to work with data in a multitude of ways supported by the design of the outsourced system. The scale of the data requires this approach because it is no longer cost-effective to maintain such systems in-house. The economies of scale (shared components, shared public data, shared features, etc.) of outsourcing to a co-tenanted platform are so much better, and are much less of a hassle/responsibility to maintain than employing staff to manage the service in-house.

Outsourced bioinformatics (note informatics, not analytics) platforms are becoming more common, including those driven by the recent Pistoia Alliance Sequence Services project. They store data and provide facilities for researchers to design and implement their own detailed queries using pretty much any analysis tool they like. They still require the users to employ experts in biostatistics, biology, biochemistry, bio* in general to be able to select the correct tools and configure input and interpret the output in an appropriate way. They are not expert systems, but informatics systems that support experts.

Outsourced analytics though are an entirely different kettle of fish. These services are characterised in two forms - the first being a set of predefined, detailed, well-engineered and effective analysis workflows that process data in a predetermined way to come to reliable, possibly even clinically-applicable/validated answers. The second is the provision of research as a service, where external experts are brought in to analyse and interpret a customer's data to find the meaning within, so that the customer does not have to hire their own expert research staff. In effect, these companies are bioinformatics CROs.

It is one thing to outsource the IT infrastructure that supports your own research, but quite another to outsource the research itself. Outsourcing the IT infrastructure does not affect your ability to develop and own unique IP and to retain an expert base of staff who really bring great value to your research pipeline. Outsourcing the analytics itself, which by default would mean no longer requiring the services of so many of your own research staff, makes you entirely vulnerable to the professionalism of an external partner who will most likely be working to deliver only precisely what was requested, no more and no less, and internally trying to do so with as little effort as possible without taking any unnecessary risks or chancing too many innovative new ideas.

There will, and should, always be a place for analytical staff in every biotech company, including pharma, who can make use of informatics platforms in-house or outsourced to come to research conclusions which their employer will own and be in complete control of. Removing these to a CRO providing outsourced analytics necessarily transforms the entire business model of the customer into a virtual pharma, commissioning and licensing ideas from third parties and providing in-house only the strategy and marketing to create profitable products out of the gathered information. This may not be a bad thing, it depends on your business model and vision, but it is not and should not be confused with outsourcing informatics!

Topics: analytics, big, Big data technology, Bioinformatics, Bioinformatics, Cloud, data, DNA, genome, genomics, outsourcing, scale