Most of my experience has been working with genomic data, this time I am working on data from Smart-seq3. RNA excreted by cells in culture medium were processed with Smart-seq3 (From the looks of it, to capture as many fragments as possible). I have to perform differential expression analysis and check for any significant genes. I have the fastq files and trimmed them for adapters and polyA using trim_galore.

The protocol suggested processing fastq file using zUMIs, however the data isn't traditional single cell sequencing, so as per one comment of their procotol, they replied that for bulk sequencing-esque applications, we can just trim out the UMIs.

I have not found a decent way to remove UMIs from fastq file, are there any tools to remove them? From what I've read, UMI_tools can extract the UMI information from the fastq file, but how I remove it from the fastq file?

  • $\begingroup$ May I ask why Smart-seq3 protocol was used if you plan not to use UMIs? I understand you will want to trim away the cell barcodes (if the RT primer contained them), as they are meaningless in bulk sequencing, but UMIs are there to enable more accurate counting of RNA molecules, especially if the amount of starting material is low, which requires a high number of PCR cycles in library preparation. $\endgroup$
    – Cloudberry
    Sep 12, 2023 at 21:03
  • $\begingroup$ More like, I have entered the project during the analysis stage, so I was not involved in designing the project, and now I have been tasked to analyze the data in the most meaningful way I personally can. edit2add - I have not worked with UMIs or RNA data. Only 30-40% of reads have UMI, so even after collapsing reads normalization and analysis might be an issue. Should I analyze UMI reads separately? $\endgroup$ Sep 14, 2023 at 11:50
  • $\begingroup$ I have never worked with Smart-seq data, but in our lab we use a Drop-seq-like protocol for multiplexing bulk RNA-seq samples, and we process the data using the Drop-seq-tools software bundle. One idea could be to just use zUMIs as recommended for single-cell data, since it can be set up to handle Smart-seq v3 reads. Once you have the differential gene expression matrix, you can then continue with regular RNA-seq analysis tools. $\endgroup$
    – Cloudberry
    Sep 14, 2023 at 18:46
  • $\begingroup$ Removing the UMIs may be trickier because, as you say, not all reads contain them. It looks like zUMIs can be set to output demultiplexed BAM files per cell (in your case per sample), so if you don't want to use UMIs, you could count the reads from those files, ignoring the UMI tags. $\endgroup$
    – Cloudberry
    Sep 14, 2023 at 18:50


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.