What to do when using Decontam on a set of samples, but a few samples did not have detectable DNA concentration values?

[Kudzai-H]

Hi @ benjjneb,

I would like to run the Decontam frequency method on 54 samples. However two of them had DNA concentrations that were not detectable on the Qubit. May I ask for your advice on whether it would be better to remove them in the decontamination process or to put a replacement value (i.e. the lowest DNA concentration that was detected across all 54 samples)? No negative controls were included.

[benjjneb]

I would remove them for the purpose of running decontam

[Kudzai-H]

Thank you very much. Would you then recommend adding them back again as they are after decontam or pruning the same contaminant taxa from them and then adding them back in after decontam? I do not suppose there are any tutorials on filtering samples prior to deontam, then adding them back before the contaminant pruning step? From: Benjamin Callahan ***@***.***> Sent: Thursday, February 6, 2025 10:05 PM To: benjjneb/decontam ***@***.***> Cc: Kudzai Hwengwere ***@***.***>; Author ***@***.***> Subject: Re: [benjjneb/decontam] What to do when using Decontam on a set of samples, but a few samples did not have detectable DNA concentration values? (Issue #159) I would remove them for the purpose of running decontam - Reply to this email directly, view it on GitHub<#159 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BOMWK7MMIGBEXOT63XSJF432OPMBDAVCNFSM6AAAAABWTHTVTSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMNBRGE3DENRTHE>. You are receiving this because you authored the thread.Message ID: ***@***.******@***.***>>

…

________________________________ [https://www.plymouth.ac.uk/images/email_footer.gif]<http://www.plymouth.ac.uk/worldclass> This email and any files with it are confidential and intended solely for the use of the recipient to whom it is addressed. If you are not the intended recipient then copying, distribution or other use of the information contained is strictly prohibited and you should not rely on it. If you have received this email in error please let the sender know immediately and delete it from your system(s). Internet emails are not necessarily secure. While we take every care, University of Plymouth accepts no responsibility for viruses and it is your responsibility to scan emails and their attachments. University of Plymouth does not accept responsibility for any changes made after it was sent. Nothing in this email or its attachments constitutes an order for goods or services unless accompanied by an official order form.

[benjjneb]

Run decontam on the 54-2=52 samples with DNA concentration information. Identify contaminants. Then remove the identified contaminants from the full table with all 54 samples.

[Kudzai-H]

Thank you. I will do just that. I apologise as I have one more question. I ran the command {hist(contamdf.freq$p, n=100)}, in order to help me decide on the threshold to use for my decontamination. The histogram I retrieved (picture attached to the left) is a bit difficult to interpret, as I seem to have abundant sequences classed as contaminants and abundant sequences that were not classed as contaminants. It seems to neither be bimodal or linear as shown in other reports. The frequencies are very low (maximum 15) and I have only three samples in some batches (I ran the decontamination batch mode) so I wonder what you think of this histogram plot. Removing the batch mode creates a slightly more linear histogram (picture attached to the right) but leads to more sequences being classed as contaminants. I would like to ask what you think of these decontam scores. ***@***.******@***.*** From: Benjamin Callahan ***@***.***> Sent: Friday, February 7, 2025 1:49 PM To: benjjneb/decontam ***@***.***> Cc: Kudzai Hwengwere ***@***.***>; Author ***@***.***> Subject: Re: [benjjneb/decontam] What to do when using Decontam on a set of samples, but a few samples did not have detectable DNA concentration values? (Issue #159) Run decontam on the 54-2=52 samples with DNA concentration information. Identify contaminants. Then remove the identified contaminants from the full table with all 54 samples. - Reply to this email directly, view it on GitHub<#159 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BOMWK7LFFA6QMRYC7DQIBXD2OS2TBAVCNFSM6AAAAABWTHTVTSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMNBSHE4TMNRRGM>. You are receiving this because you authored the thread.Message ID: ***@***.******@***.***>>

…

________________________________ [https://www.plymouth.ac.uk/images/email_footer.gif]<http://www.plymouth.ac.uk/worldclass> This email and any files with it are confidential and intended solely for the use of the recipient to whom it is addressed. If you are not the intended recipient then copying, distribution or other use of the information contained is strictly prohibited and you should not rely on it. If you have received this email in error please let the sender know immediately and delete it from your system(s). Internet emails are not necessarily secure. While we take every care, University of Plymouth accepts no responsibility for viruses and it is your responsibility to scan emails and their attachments. University of Plymouth does not accept responsibility for any changes made after it was sent. Nothing in this email or its attachments constitutes an order for goods or services unless accompanied by an official order form.

[benjjneb]

There is no picture attached to your post.

It looks like you are posting via email, you might get better results going to the Github issues page and using that interface to post to this thread.

[Kudzai-H]

Apologies for that. I have attached them here.

Hopefully you can see them now.

[benjjneb]

The plots display now, but which plot is to the "left" or "right" is still unclear.

[Kudzai-H]

Sorry again, the first one with a frequency scale going up to 30 is right and the second one with a frequency scale reaching 15 is left.

[benjjneb]

Running in non-batched mode looks preferable based on this distribution of scores.

Like you noted, you are not seeing the idealized bimodal distribution of scores that is typical of a clean mixture of contaminant and non-contaminants. This could be for a variety of reasons (the data is mostly contaminants, your real samples have little overlap with one another, there is too little variation in DNA concentrations to effectively distinguish, ...).

If you need to use a score threshold to remove contaminants, I would use taxonomic inspection differently scored ASVs to get a sense of the plausibility as true sequences at different score levels. I would then consider whether it is more important to keep contaminants out of my analysis, or to keep in real ASVs and just remove the most egregious contaminants. Finally I would also plot a read-weighted histogram to see if that shows a clearer signal.

But what I might do instead is to not remove contaminants at all, but to use the decontam scores to interpret the downstream results of my analysis. For example, if I do a differential abundance analysis that flags ASVX -- what is the decontam score for ASVX? If it is a low (contaminant-like) score, I would down-weight its importance in my interpretation.

An example of this second approach is in our recent paper looking at feline bile: https://doi.org/10.1111/jvim.16852

[Kudzai-H]

Thank you very much for this advice!