[Feature request] Custom cell and UMI barcode position

[Hoohm]

Hello,

I would like to use alevin but I'm not sure I actually can. We use a custom R1 barcode. 1-7 for the cell barcode and 8-16 for UMI.

Is is complicated to implement this?

[k3yavi]

Hi @Hoohm ,
Thanks for the feature request.
Currently Alevin do have a hidden feature, where you can explicitly specify the CB and UMI length. Although we have not yet extensively tested these options but in your settings you might have to specify the following command line argument:

--barcodeLength 7 --umiLength 9 --end 5

Please let us know how it works out for you in these settings, it will help validate these options for Alevin.

PS: Just a quick question for my understanding, is there a specific reason you chose to use the length of the UMI longer than CB in your experiment ?

[Hoohm]

Hi @k3yavi ,
Thanks for the info!

We are working on an optimized version of SCRBseq and one of the problems we had with the original protocol is the minimum distance between the cell barcodes being too low. So we increased the number of bases. The original protocol was 6 bc and 10 umi. We just switched the 7 position from umi to barcode.

We use a known whitelist of barcodes since it's a well plate based protocol. We know that any other barcode are not cells. Is there an option for max distance allowed between BC or UMI?

[k3yavi]

Hi @Hoohm ,
Thanks for the quick reply and the explanation .

I personally am not very well versed in the working of SCRBseq. But, as you explained, knowing a set of whitelist CB beforehand is always a plus for the downstream working of the pipeline. Currently, Alevin merge all the observed CB which are 1-edit distance from a known whitelist CB towards the whitelist. The underlying assumption being that the sequencing error (although with low probability) can change CB sequence and we can correct for that. I wonder, is this right to do for your experiment ?

re: >Is there an option for max distance allowed between BC or UMI?
Sorry, but I don't completely understand this question. When you say distance allowed between CB and UMI, do you mean there is a sequence between CB and UMI ( like in in-drop seq)? If that's the case then we might have to tweak a bit in alevin command line flags again.

But I suspect what you meant by above statement is -- max distance allowed between CB among themselves. If that's the question then unfortunately we currently allow correcting for only 1-edit distance for both CB and UMI. But if you think more correction is needed by your protocol then we can put this on the feature request list and discuss about working on this on the next release.

[Hoohm]

Hey @k3yavi , Thanks for the fast answer.

You got it right. I was asking about the max distance allowed for cell barcodes and UMIs.

I will try it out with the default options and come back to you for a potential future feature request.

[k3yavi]

Hey @Hoohm , we have released v0.11.1 with some fix for the customized length mode.

Let us know how it worked out for you.

[k3yavi]

@Hoohm , I believe the custom length options looks good from our end. Feel free to reopen the issue if you face any problem while using alevin in this mode.

Re: More customizable options like a regex for extracting CB and UMI is still in development and has been raised in issue #233 and will keep that issue open until we have more generic extraction.

Thanks again for the feedback and useful comments.

[kh49]

How do I apply the custom length settings if I am utilizing the wrapper for v1 10x data? I have v1 data that has a 5bp instead of 10bp UMI but everything else is the same. Can i just add --umiLength 5?
Data:
https://support.10xgenomics.com/single-cell-gene-expression/datasets/1.1.0/cd14_monocytes

[k3yavi]

Hi @pophipi ,
I think if you specify --gemcode along with all three i.e. --end 5 --umiLength 5 --barcodeLength 14 as the command line flag it should work for you.