By Sramana Mitra guest author Shaloo Shalini
SM: What you are saying here is bio medical researchers would benefit from solutions, SaaS maybe, which can take the hard technical side out and enable them to focus on what they are trained to do.
Let me ask you a follow-up question. If you were to give some guidance to entrepreneurs who come from the software service world and want to help provide that infrastructure to the bio medical research community, where would you suggest they begin that exploration? What fields of biomedical science are most active at the cusp of computational modeling and biosciences?
MA: There are probably three areas that I would point entrepreneurs to. One would be this entire industry of sequence analysis and the next-generation sequence analysis. In this genetic sequencing field, there are definitely companies that are focused on this to some degree, but I don’t know if there are any good offers in an SaaS model tying some of these activities into public or private clouds.
Number two would be image analysis tools. A lot of time is spent getting these high-resolution digital images and then managing them, providing tools to serve all the classic stuff you must do with these images such as find edges, find patterns, and find other things related to medical research. Obviously there is software out there such as MatLab image analysis tools catering to all these requirements. But these current tools require a fair amount of technical expertise to run. If there is a certain categories of image management that is best suited to SaaS, that is another open opportunities.
Finally, there is modeling and simulation. I think there are some meta languages around modeling and simulation, so you could see if we get the medical researcher’s stuff in a common format so that they can use SaaS approach for sharing and collaboration. These are the big three areas where I expect to see innovation and new cloud-based solutions from entrepreneurs.
[Note to readers: Research projects such as Microsoft Azurescope and its Generic Worker framework aim to address some of these areas for Windows-based bioinformatics and medical research environments. Instead of requiring science users to write code to deploy their existing desktop applications to Azure, it aims to provide simple tools that register .NET, Java, MatLab and command-line applications to “Generic” workers in Azure. Usually, activating the remote cloud applications from desktop clients like workflows requires additional code for parameter passing through queues and file transfer between desktop and cloud storage. This framework provides tools for transparent, on-demand invocation of registered cloud applications without writing code from command line, workflows, or APIs to bridge the gap between client and the cloud. You can read more about Azurescope here and Azureblast in an excellent paper here.]
SM: Very good pointers. What is the status of the biomedical research industry across United States and the world in terms of researchers being able to do the kind of research that you are doing? You clearly have a level of infrastructure already put together that is expensive and hard to manage. What does the hierarchy of a typical research institution or lab look like where that kind of infrastructure exists? What would it take if you were to allow a larger set of researchers across different universities and across the world to be able to leverage that infrastructure?
Say you have today a private cloud; if you were to give some access to some of that data to researchers who don’t have that kind of infrastructure in some lower tier institutes, let’s say in China or Malaysia. What is the lay of the land in terms of research institutes and collaborations globally?
MA: That is an excellent question because there is nascent activity in this regard. The bottom line right now is informatics activities in biomedical research are very disorganized. It is hard to collaborate, and a lot of activities are traditionally done in silos. Now, part of this is a bit of a natural occurrence when you are dealing with, say, patient data. You have to obviously have to cover such patient data to take care of privacy laws.
But there are other models in the computational research space, for example high-energy physics, that manage to have a huge community around the world sharing data and working together. Part of the reason is things such as there are only a few mass particle accelerators in the world, one in Geneva and a few more at other locations. So, a lot of people have to collaborate because they are using same instruments and things. But they have developed a lot of tools along the way which added security mechanisms to account for patient data, even setting aside patient data. But that is only basic type of research data I have been talking about.
There is an opportunity in biomedical research to really bring together a community just as in high energy physics and other places. There are activities already starting. One piece that we have solved here for biomedical research only is the federated authentication and access control mechanism. There is a group called Incommon that has most of major universities on board by using a shibboleth-based authentication mechanism that allows people to use their home institution credentials for collaboration and research, and they lose access to remote servers and their home instituted if their credentials become invalid, if, say, they get fired for embezzling funds or something like that. This piece is in the works.
There are also a number of different consortia that are trying to address such requirements, including one group that we are putting together in New England called the biomedical information technology consortium. It is meant to bring together the leaders of research IT activities that most of the major academics and pharmaceutical operations in New England who have started to figure out how we can move toward a collaborative shared computing environment.
[Note to readers: You may want to read more on some early adoptions and how cloud computing is helping scientists run high-energy physics experiments here.]
SM: That is very interesting to know.