This is a test version of Biostars. For the public version, visit https://www.biostars.org.
The bioinformatics paper reappears, but the results are significantly different from the original results

The bioinformatics paper reappears, but the results are significantly different from the original results. I'm a newbie and just started doing bioinformatics analysis, so there's a lot I don't know. The following is my entire process, but the result is very different from the original article. I would like to ask here, is there something I did not do well or is there something wrong? I hope you seniors can give me some opinions and suggestions!

Here, the analysis of RNA-seq and DAP-seq data from this article is replicated. The data files and code are available in GitHub.

reproducibility
• 2,078 views • link
written by yuanj • yuanj

The joys of trying to replicate reported results. I've never managed to replicate a result 100%. Usually, it's an acceptable level of variation, or down to vague methods sections that makes it impossible to fully understand what parameters they used.

But it sounds like you've tried and gotten something very different. It's impossible to understand since you haven't shown what they've got and what you've got. Talk to your PI, if it's an important paper and others can't replicate their results at all, then they can make the call to contact the journal.

• 0 views • link
written by dthorbur dthorbur

The original article is here and my process is here.I have contacted the corresponding author and hope that the problem can be solved.

• 0 views • link
written by yuanj • yuanj

I agree with Basti, and reiterate that you need to show how the results differ as we cannot make comment on the size of the difference without assessing it ourselves.

• 0 views • link
written by dthorbur dthorbur

Different pipelines will inherently yield different results but what you mean with "very different results" need be to clarified. There is no information on the alignment process used in the article so this is a first discrepancy that could yield some variation with your results. But above all, you applied DESeq2 while they used DEGseq, two packages that rely on different assumptions and statistical methods to identify DEGs, so it is certain it will yield different DEGs. As the process given in the article is impossible to 100% reproduce, are your results still in accordance with the results obtained with your method ? Maybe the number of DEGs is highly different but the main conclusions (top DEGs, main pathways modified) may be the same ?

• 0 views • link
written by Basti • Basti

1 answer

Mainly the number of differential genes in RNA-seq analysis and the consistency of sample peaks in DAP-seq experiments are different from the original results.

After communicating with the author of the original article, it was found that this difference may be caused by differences in analysis methods and filtering parameter settings (including differences in the analysis software used). The original article may have had more stringent screening criteria applied. It is a common phenomenon that different analysis methods and screening criteria lead to differences in the number of genes. Nonetheless, the reproduced results still support the discussion in the original article.

I'm a newbie and may not know the best way to ask questions. I'm sorry to take up your time. I will correct some of my practices in work and questioning.

• 0 views • link
written by yuanj • yuanj

Log in to answer this question.

More posts like this