Building a splici index for two different read lengths

[lawsonvt]

Hello,

I have a pre-processed, paired-end snRNASeq dataset I want to to align in which the first read is 26bp and the second read is 91bp. How do I generate a single splici index for two different read lengths?

Thank you,

~ Mark

[rob-p]

Hi @lawsonvt,

In this case, the first read is techinal (UMI and barcode) and not relevant to the index construction. You should treat 91 as your read length.

Best,
Rob

[lawsonvt]

Thanks Rob. Does this mean the reads should be aligned as single reads or still as paired?

[rob-p]

What protocol were they sequenced with? salmon alevin will accept a pair and just "do the right thing" if you pass in the correct protocol flag.

Best,
Rob

[lawsonvt]

"snRNA-seq libraries were obtained using 10X Genomics Chromium Single Cell 5′ v2 chemistry following nuclear dissociation"

Is that enough info?

[rob-p]

Yes; this is sufficient information. Basically, it means you should run salmon alevin with the --chromium flag — with something like:

salmon alevin quant -i <index> -l ISF -1 <read1> -2 <read2> --chromium --sketch -p <num_threads> -o <output_map_dir> 

Then, when you perform the generate-permit-list step of alevin-fry, make sure to filter with the orientation one expects for the 5' 10x protocol. That is, you should pass -d rc to the generate-permit-list command, since we expect the biological read to map in the reverse complement orientation in this protocol. Otherwise, the process should be as normal. Let us know if you run into any other questions.

--Rob

[robertzeibich]

I ran into something similar. I have FASTQs with different read lengths:
R1: 26 R2: 124
R1: 150 R2: 150
R1: 151 R2: 151

Do I have to create different reference indices?
What should r-len be? Should I use 91, or 124, 150, and 151 instead?

I am using simpleaf (alevin-fry).

[robertzeibich]

I just had a look at the following post. Since the specific length does not seem to make a difference, I will use 91. If I am mistaken, please correct me.

Based on the following tutorial, you recommend checking and using the exact read length of the processed dataset.

I am still confused.

A response would be much appreciated.