New research outlined in a paper published in the journal Science describes the results of a collaboration between Kioxia Corporation and MIT’s Synthetic Neurobiology group, led by Edward Boyden. The research, involving both MIT and Harvard scientists, was conducted while members of Kioxia and Boyden’s group were working together at the MIT Media Lab, and presents a novel way to locate and sequence RNA within tissue samples—potentially opening avenues for researchers to learn more about gene expression in cells.

The beginnings of this work occurred as early as 2014, when a team led by George Church at Harvard Medical School developed an RNA sequencing technique known as FISSEQ (fluorescent in situ sequencing); tangentially, the Boyden lab had innovated the expansion microscopy technique in 2015, and both Church and Boyden’s groups were looking for a way to combine the two methods and produce higher resolution in situ sequencing results using tissue expansion. The new procedure born from this joint effort is called expansion sequencing, or ExSeq.

Kioxia, a Media Lab member company, also began to collaborate with Boyden’s lab in 2015, and initially focused on projects relating to microscopy image preprocessing. However, while stationed in Cambridge, MA, visiting scientist Dr. Yosuke Bando (Bandy) saw an opportunity for Kioxia's computer memory technology to contribute towards this life science research in a new way: data storage had become a central issue in the expansion sequencing project, as Synthetic Neurobiology researchers estimated that petabytes of data would have to be stored. The researchers needed to scale up their existing technology—which processed data at a very slow rate—to analyze a growing volume of sample data.

In order to aid this acceleration, Kioxia built a GPU cluster equipped with solid-state drives (SSDs) and developed image analysis algorithms, which the researchers applied to the ExSeq project. Consequently, the multi-day data processing time was reduced to a few hours, which meant that the researchers could obtain the results of their experiments much more quickly and expedite the whole method pipeline, from sample preparation to imaging to data processing. 

“The researchers were therefore able to analyze 10 terabytes of data from a large specimen in just a few days,” said Shigeo “Jeff” Ohshima, Technology Executive at Kioxia. “It is said we saw two years' worth of digital transformation in two months...it may not just be a coincidence that we made the researchers' lives easier by the equivalent velocity.”

It’s important to note that this general process for data storage was iterative and continued until late 2019; Boyden lab researchers kept improving the protocol by looking at the analysis results, which required some modifications to the data processing and further acceleration of the modified computer code. The overall positive results of this collaboration, though, reflect the serendipitous nature of the research conducted at the Media Lab—how seemingly disparate fields can combine in unexpected ways to achieve great work, while simultaneously learning from and building off one another. 

"We worked with Kioxia to analyze these enormous datasets to extract the patterns found within,” Boyden said. “This was a true collaboration, joining forces to make new science possible. We found patterns of gene expression in brain cells, as well as in tumor biopsies, that may point to new ways of devising therapeutic directions in the future."

The teamwork demonstrated by members of Kioxia and Boyden’s lab further highlights the myriad of skill sets that can be found within the Media Lab community. For example, Boyden’s team provided technical knowledge of neuroscience and life science, while Kioxia’s data storage products and industry knowledge became an invaluable part of the project. The fact that Boyden’s researchers were highly skilled in both computer science and biology allowed them to patiently explain the research in layman’s terms, so members of Kioxia could understand and gradually realize how the company might contribute towards the project. 

“Through this exciting collaboration with Professor Boyden and his talented team, we have recognized how our flash technologies can support life science,” said Ohshima. “We are beginning to expand our reach into the life science community and are thrilled to work with [MIT] further as well.”

***

This research is featured in an article written by MIT News.

Read the press release in English on the Kioxia website.