Researchers at Kyoto University’s Institute for the Advanced Study of Human Biology (WPI-ASHBi) have developed a statistical method that improves the quality of single-cell datasets across multiple technologies. The approach, called iRECODE, extends an earlier method known as RECODE by addressing both technical and batch noise in high-dimensional single-cell data.
Single-cell RNA sequencing is powerful for mapping gene expression in individual cells, yet measurement dropouts and experimental inconsistencies can obscure signals. RECODE, introduced in Life Science Alliance in 2022, helped address technical noise by using high-dimensional statistical modeling to correct expression distributions without complex parameters or machine learning. iRECODE builds on this foundation by also correcting for batch variability, allowing researchers to more reliably compare data across experiments.
When applied to single-cell RNA-seq, iRECODE clarified gene expression profiles, reduced sparsity, and improved cell-type alignment across batches while preserving cell identity. The method proved efficient, operating about ten times faster than using separate noise reduction and batch correction pipelines. It also performed consistently across protocols such as Drop-seq, Smart-seq, and 10x Genomics platforms.
The research team demonstrated that iRECODE works with more than just RNA-seq. In scHi-C datasets, the method reduced sparsity and highlighted meaningful chromosomal interactions. In spatial transcriptomics, it clarified tissue-level gene expression patterns across different platforms and species.
“Single-cell data captures countless cellular ‘whispers,’ but hearing those whispers through the noise is extremely difficult,” said Yusuke Imoto, who led the work. “iRECODE, an evolution of our RECODE method, lifts those voices to the surface. Through this method, I believe the hidden stories of cells, stories we could never hear before, will steadily come to light."
Publication Details
Imoto Y. Comprehensive noise reduction in single-cell data with the RECODE platform. Cell Rep Methods. 2025;5(9):101178. doi:10.1016/j.crmeth.2025.101178
Single-cell RNA sequencing is powerful for mapping gene expression in individual cells, yet measurement dropouts and experimental inconsistencies can obscure signals. RECODE, introduced in Life Science Alliance in 2022, helped address technical noise by using high-dimensional statistical modeling to correct expression distributions without complex parameters or machine learning. iRECODE builds on this foundation by also correcting for batch variability, allowing researchers to more reliably compare data across experiments.
When applied to single-cell RNA-seq, iRECODE clarified gene expression profiles, reduced sparsity, and improved cell-type alignment across batches while preserving cell identity. The method proved efficient, operating about ten times faster than using separate noise reduction and batch correction pipelines. It also performed consistently across protocols such as Drop-seq, Smart-seq, and 10x Genomics platforms.
The research team demonstrated that iRECODE works with more than just RNA-seq. In scHi-C datasets, the method reduced sparsity and highlighted meaningful chromosomal interactions. In spatial transcriptomics, it clarified tissue-level gene expression patterns across different platforms and species.
“Single-cell data captures countless cellular ‘whispers,’ but hearing those whispers through the noise is extremely difficult,” said Yusuke Imoto, who led the work. “iRECODE, an evolution of our RECODE method, lifts those voices to the surface. Through this method, I believe the hidden stories of cells, stories we could never hear before, will steadily come to light."
Publication Details
Imoto Y. Comprehensive noise reduction in single-cell data with the RECODE platform. Cell Rep Methods. 2025;5(9):101178. doi:10.1016/j.crmeth.2025.101178