Clearing the fog, new method brings more realistic single-cell RNA sequencing data

Knowing the profile of RNA in a cell can show which genes are active, allowing researchers to speculate about what the cell is doing

However, RNA quantification in microsamples also poses significant technical challenges

Recently, a research team at the Institute for Advanced Study in Human Biology at Kyoto University has developed a new mathematical method that can remove noise and help people extract clear signals from single-cell RNA-sequencing data

"Each gene represents a different dimension in the RNA-sequencing data, which means tens of thousands of dimensions must be collected and analyzed," explains Yusuke Imoto, the paper's lead author

To break the curse of dimensionality (COD), the research team developed a new noise reduction method, RECODE, to remove random sampling noise from single-cell RNA-sequencing data

First, the researchers tested their method on extensively studied human peripheral blood data

Second, RECODE has the advantage of being a more realistic representation of gene activation when compared to other advanced analysis methods

Finally, the research team tested RECODE on a complex dataset of cells from mouse embryos, which contains many different types of cells with unique gene expression patterns

Imoto concluded: "Data analysis of single-cell RNA-sequencing remains technically challenging, and it is an evolving technology, and our RECODE algorithm is a big step toward uncovering the true behavior of individual cells

Another lead author, Tomonori Nakamura of Kyoto University, added: "By unleashing the true power of single-cell RNA sequencing, RECODE will help researchers discover unknown rare cell types, develop and establish in basic research as well as clinical applications and drug discovery.

The RECODE calculation program (Python/R code) is available from the GitHub (https://github.

Original text retrieval

Yusuke Imoto, Tomonori Nakamura, Emerson G.