This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Unable to validate biomarkers obtained from single-cell RNA-seq data

I have compiled a list of biomarkers related to chemoresistance in triple-negative breast cancer patients using single-cell RNA-seq data. I am now performing validation on independent patient cohorts with microarray measurements. Some of the top genes identified by my method include a lot of single nuclear RNA (snRNA) genes such as SNAR-A10, SNAR-A11, SNAR-A7, SNAR-B8, etc. I cannot find these genes in the microarray dataset and am stuck at the validation step. My questions are as follows:

  1. Should I remove these RNA genes upfront and continue with the analysis? That way, these won't show up as biomarkers.
  2. Is there any way to map these genes to the microarray data (GPL96 HG-U133A Affymetrix Human Genome U133A Array)?

Any help is highly appreciated.

biomarkers single-cell seurat
• 1,289 views • link
updated by ATpoint ATpoint • written by Gene_MMP8 Gene_MMP8

1 answer

These short RNA species (snoRNA, miRNA etc) are usually not covered in arrays or RNA-seq since the array or library preparation as well as initial RNA extraction do not capture them well. I am not sure whether scRNA-seq should capture them or whether this is rather noise, check the literature. As for your problem, you need dedicated smallRNA-seq datasets or arrays for this analysis.

• 0 views • link
written by ATpoint ATpoint

Thank you for your comment. I will look through the literature and try to find dedicated arrays. Would you suggest removing these genes from the single-cell data? I am thinking in lines of preprocessing.

• 0 views • link
written by Gene_MMP8 Gene_MMP8

I would consider that indeed. Maybe remove all small RNA types or retain only biotypes such as lncRNAs and protein-coding genes. This can be fone directly be filtering the count matrix.

• 0 views • link
written by ATpoint ATpoint

Log in to answer this question.

More posts like this