So basically I have RNA-seq reads that were generated from Illumina and Ion Torrent platforms for yeast species. I have seen an article where they compared liver cells of a rat that were sequenced with these two platforms, but they made use of PORT bioinformatics tool for the normalization of the data. This requires input data to be in the form of the Ensembl database. This database only has Saccharomyces cerevisiae annotation files available on it, whereas I have other species that I need annotation files for as well, which are available in RefSeq format from NCBI.

So, I am still brand new to doing RNA-expression analyses and have never done anything remotely at this scale. Please could anyone give me advice on the steps involved in analyzing the paired-end and single-end reads together as comparable data sets and then carrying out a complete differential expression analysis on this data. Any advice or guidance would really be appreciated, I am honestly not sure how to approach this at all.

  • 1
    $\begingroup$ link to the various studies? Is single-end IonTorrent only or is there SE data for Illumina too? Why do you have both read types? Are you trying to compare the two technologies, or are you trying to compare different samples with the only available data which happens to be IonTorrent in some samples and Illumina in others? I worry about comparing results confounded by technology, as this could lead to misleading conclusions,. $\endgroup$ Sep 21, 2021 at 20:36
  • $\begingroup$ Hi there @MaximilianPress, thank you for your answer. Only 1 of the datasets is publicly available, it is one of the paired-end libraries generated by Illumina. The idea is to compare the samples from the different studies. There is only single-end available from IonTorrent sequenced samples and only paired-end for the Illumina sequenced samples. I spoke with my supervisors though, they said to just compare the counts data for each independent study to each other. Starting with a PCA plot of the data from the different studies. $\endgroup$
    – Justin1609
    Sep 25, 2021 at 13:48
  • $\begingroup$ I worry about comparing across studies with different technologies, there are likely to be significantly different sample prep, instrument, etc. issues. Sometimes it makes sense to compare across studies in this fashion but usually you want to explicitly control in some way. PCA is likely to just show that the different studies/technology explain most variation, rather than biologically meaningful variables. Again, it is hard to say much without more information. $\endgroup$ Sep 26, 2021 at 17:18
  • $\begingroup$ @MaximilianPress so basically the data is assessing the effects of physical contact between cells of different species versus when these species are grown as mono-cultures. So we are interested in the genes that are differentially expressed as a result of contact with, or the presence of, the other species. I am not sure if this makes any more sense, but my supervisors have also said there obviously will be differences in how these datasets were prepared and so forth. But, they maintain that you should still be able to detect biologically significant information from these comparisons. $\endgroup$
    – Justin1609
    Sep 26, 2021 at 17:34
  • $\begingroup$ Fair enough, I can't really provide guidance here as it sounds like it will necessarily be a little complex and I'm not sure that I understand how the different datasets will be used. Possibly update your question with some of this information? $\endgroup$ Sep 26, 2021 at 21:44


Your Answer

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