I understand that after running umi_tools, if one want to do quantification, it is possible to first map the (usually) R2 fastq file. Unfortunately, the R1 file becomes empty after the extraction, as the BC and UMIs are being moved to the header.
According to the umi tools manual, one must next run STAR to map the samples in order to count using umi_tools count.
I would like to know, if there is a way to keep the sequences in the R1 file, so that it would be possible to work with cellranger count after the barcodes were extracted.
thanks
1 answer
I'm not quite sure what you are trying to achieve.
The standard UMI-tools pipeline is umi_tools extract > STAR > umi_tools count.
Although, if you are analysing 10X chromium data, then we'd recommend alevin or alevin fry.
As Arup Ghosh says, if you could rebuild the R2 from the read headers, but why? At this point you'd achieve exactly the same as just feeding the fastq files into cellranger without running UMI-tools. The really the only parts of UMI-tools that distinguish it from any other tool are dedup/group/count. extract is just a simple string manipulation utility that in the case of 10X data does exactly what cellranger does to its fastqs anyway.
If you have non-10X that you'd like to analyse through the 10X pipeline, then as Arup Ghosh says, you could rebuild R1s from the fastq headers, but you'd also need a way to feed your BC whitelist into cellranger and i'm not sure if there is a way to do that.
I believe that STAR solo can be configured to produce an output very similar to cellranger so that might be a better way forward.
Log in to answer this question.
-
Can I count UMI-barcode combination without mapping?
written by Assa Yeroslaviz
-
UMI-Tools knee-method has great influence on the results of white list
written by Assa Yeroslaviz
-
UMI-tools no tsv files at deduplication
written by Sarah •
-
Removing umi split off as a separate fastq (RNA-seq)
written by YJ •
-
UMI extraction from 10X visium spatial transcriptome data
written by Omics data mining
-
How to combine UMIs split across reads 1 and 2 into a single UMI for deduplication?
written by SethG •
-
umi_tools using regex greedy quantifiers only returning 1 character
written by williamtmills •
-
Transferring UMI from paired-end read 2 to header of read 1
written by robert.gnuegge •
-
Tutorial: scRNA-seq CITE-seq-count bioinformatics
written by colindaven
-
Tool to append Illumina Read 1 to Read 2 for downstream demultiplexing
written by achamess
One method will be to parse the R2 fastq headers and create an R1 file with the cell barcodes and UMIs with dummy quality scores, i.e. I, J or K.
If you have the original data, you could always make a copy of the data before running
umi_tools?What exactly are you trying to do? Run non-10x data through cellranger?
thanks for all the comments, no, I have 10x Data sequenced on Illumina Nova-Seq machine. The thing was, that the user working with this workflow wanted to run the standard cellranger workflow
mkfastq->count, but we figured out that there were way too many cells for what we loaded. Loaded were 15.000-20.000 cells butcellranger countidentified around 60-80K per sample.So we are trying to understand why we get too many and now try to exculde some assumptions, such as the wrong whitelist was used, we have the wrong barcodes, too many PCR duplications, index hoping, too much ambient RNA or empty droplets
In the web summary cellranger count provides, the numbers for median UMI counts per cell or Genes per cell are very very low, so we try to correct for that. This is why we try
umi_tools extract. But yes, I admit, I have forgotten, it continues differently after extracting the correct barcodes.So I'll test it as suggested with STAR and
umi_tools deduporcount.