Stem cells and open clouds

Today, Intel, HP, and Yahoo! publically announced an expansion of our joint effort to accelerate innovation in the area of cloud computing, called “Open Cirrus.” We are very excited that four new members have joined this collaboration, and I think this provides a good opportunity to step back and understand why we at Intel Labs are investing in the project, and how this cloud research could impact people in the future.

Open Cirrus is, in essence, a giant tool to facilitate research. Each member provides a cluster of at least 1000 microprocessor cores, which are in turn networked to each other to form a worldwide cloud. Because of its open nature, this global testbed allows us to collectively research system and software innovations that make clouds work better. It also allows us to start working through a variety of issues related to operating clouds, running cloud applications, and sharing data over international boundaries. But, perhaps the most interesting part is that Open Cirrus allows us to provide a massive computing resource to researchers in a diverse array of fields.

I would like to highlight one such project that we briefly mentioned in the press release today, a collaboration between Intel and Carnegie Mellon Univerisity that could facilitate new developments in tissue engineering, regenerative medicine, and drug discovery. In turn, this could help lead to advancements in the ability to re-grow injured limbs or other body parts. It involves combining stem cell research, computer vision, and the computational power provided by Open Cirrus.

This research involves what are called “somatic stem cells” taken from adults. All adults have stem cells – they have been found in bone marrow, skin, skeletal muscles, teeth and several other tissues. Their function in the body is to maintain and heal the tissues in which they are found. However, the amount of stems cells found in adults is small, which leads to the need to find manufacturing processes to turn a few cells into large cultures sufficient for research and medical use.

An important concept associated with stem cells is what is called “stemness,” which essentially means the ability for a given cell to consistently differentiate, or transform, into other cells. The trick is to control how this transformation happens. An important factor that scientists use to achieve this control is the application of hormones, chemicals that can influence cell behavior.

But understanding how various hormones and other growth factors trigger what kind of behaviors in different candidate cells can be extremely labor intensive. Carnegie Mellon University has a process where they “print” a pattern of hormones that interact with different cells. Until recently, monitoring the growth behaviors of the various cells was a manual process, requiring hundreds of man hours to analyze the effects of just one pattern. Listen to our Future Lab Radio podcast on this topic to learn more.

This is where Open Cirrus comes into play. Working with Mei Chen from Intel Labs Pittsburgh, the researchers are using microscope-mounted cameras and time-lapse photography to monitor the cell growth instead of human eyes. The challenge is that accurately monitoring and tracking individual cells as they quickly develop in a large, constantly shifting population requires both cutting edge algorithms and a large amount of computation. The Open Cirrus testbed at Intel Labs Pittsburgh provides such a platform for massive cloud based computation, distributing the captured images to many processors to be analyzed in parallel.

This does two things: first, the computer can watch the cells continuously, potentially tracking much more information than a human could. For instance, a single cell in a large population can be tracked over a long period of time. As cells divide, the software can track their genealogy so that researchers can know which cells are siblings, cousins, or great-great-grandchildren of each other. In short, this gives the scientists much more detailed information about what happened with the cells. You can see the results in this short video clip. On the left, you’ll see the tagged cells. On the right you can see a graph of the cells’ positisons over time (the vertical axis).

The second benefit is that the scientists can dedicate more time to deriving meaning from these results rather than spending much of their time manually capturing them. Furthermore, it is a scalable solution – one could significantly increase the number of samples studied without having to bear the cost of hiring more researchers to monitor them.

I’d like to wrap up with a comment from Prof. Takeo Kanade, who is leading the research from the university side. According to him, the ability to precisely track thousands or millions of cells in real-time will open the door for even broader applications. In an email about the project, he emphasized that “executing it on the Open Cirrus cloud computing testbed has enabled us to realize a new usage model of computer vision that can make critical impact on the biological sciences.”

In the end, this is what Open Cirrus is about — not just creating a better cloud system, but catalyzing innovation across a wide variety of fields by proving a fundamentally better way to capture and analyze a wealth of data.

Comments are closed.