AI analysis of submicron-resolution spatial transcriptomics
A new method, developed by researchers, creates images that are worth many gigabytes of data, which could revolutionize the way biologists study gene expression. Seq-Scope was first described in Cell in 2021 as the first method to analyze gene expression at sub micrometer-scale spatial resolution.
The team has since improved Seq-Scope, making it more versatile, scalable, and accessible, which was just published in Nature Protocols. Additionally, the same group has developed an algorithm for analyzing high-resolution spatial data from Seq-Scope and other technologies, called FICTURE, described in Nature Methods.
“Basically, we are hacking DNA sequencing machines and letting them do all of the hard work,” said the senior author.
Researchers use these machines to produce readouts of the transcriptome, the collection of all RNA transcribed from genes with a given cell or tissue. Traditionally, biologists studying genes within a cell or tissue must contend with the fact that a transcriptome has tens of thousands or more genes expressed, too much to make heads or tails of without the help of a computer when it also involves millions of cells.
“The problem is traditionally, there are no computational methods that allow us to understand this data set at microscopic resolution,” said another co-corresponding author.
The authors proof-of-concept method, Seq-Scope demonstrated that a sequencing machine can be repurposed to profile spatially resolved transcriptomes, enabling scientists to see how and where a gene is expressed at microscopic resolution. The team subsequently has made Seq-Scope even more cost effective, reducing the cost of high-resolution spatial transcriptome profiling from upwards of $10,000 to around just $500.
Furthermore, the new FICTURE method enables investigators to analyze massive amounts of data, by pooling the surrounding data together to make a more accurate inference at the micrometer level. By doing so, they demonstrate, you can see where cell transcripts are located without any bias.
The method generates incredibly detailed images of tissues and cells from its microscopic resolution analysis.
For example, with traditional analysis, “even if you have cell segmentation, if you don’t know exactly which cells are being transcribed and stained, the analysis can be misleading or unclear,” said the author.
“Using FICTURE, for example, you can see that skeletal muscle tissue from a developing mouse embryo is differentiating into long striated muscle cells from myoblasts.”
“We’re getting a lot of emails from companies and other investigators who previously assumed they wouldn’t be able to do such experiments and analyses. Now they are in the realm of possibility,” said the author.
The authors next hope to develop a way to make the method even more accessible to researchers, enabling them to study genomic expression from beginning to end.
