Meta transcriptomics

[Skumarerva]

Good afternoon, I am trying to run a meta transcriptomic analysis for low biomass samples, am interested in applying the decontam package to remove decontaminants after running Kraken . My question is that would you recommend to use decontam package to remove the contaminant reads rnaseq experiments. I have asked the sequencing core to provide the DNA content, which I don’t think they have. They have provided the RNA content.

I apologize for these questions, I do want to be careful in interpreting the results from rnaseq given the recent criticisms for the methodology especially for low biomass samples.

[benjjneb]

How are you proposing to use decontam on this data? Do you have negative controls? Or are you intending to use DNA (or RNA?) concentration data with the decontam-frequency method?

[Skumarerva]

Hello, I apologize for the delay in response, I didn't the answer as I was waiting for the collaborator to respond.They finally responded and said that do not have negative controls. I am an analyst and primarily work with gut microbiome . The collaborators have provided me with the RNA content so I was planning to use that, however , reading prior comments on your repo, I am a bit concerned. As i do not want to run my pipeline without effective QC measures. Additionally, the authors of this manuscript and they said they used DNA content to run decontam.
https://pmc.ncbi.nlm.nih.gov/articles/PMC10913719/
My question is , in the absence of negative controls, can I use RNA content to use decontam?

[benjjneb]

Apologies for the non-answer, but given that I have no experience working with this kind of data myself, I don't feel comfortable making a statement about whether decontam-frequency with RNA quantitation data will work effectively. This depends on some key details, such as how RNA is being quantified, how RNA is being prepared into the sequencing libraries, and how taxa abundances are being calculated, that I don't fully understand in this context.

If you wanted to try decontam in this way it would probably be best to do it in an exploratory fashion at first. Probing the score distribution (is it bimodal low-score/high-score as per decontam's presumed mixture model?), and the identities of low-score taxa (do they "look like" contaminants) and high-score taxa.