How to apply a set of 8 fastq files produced by Paired-end NovaSeq to bowtie2 or bwa for alignment?

Question

[Chilly]

I performed 10x Chromium single cell seq and used Paired-end NovaSeq to generate a set of 8 fastq files for each sample.

About filename rules:

[Sample Name]_S1_L00[Lane Number]_[Read Type]_001.fastq.gz

Where Read Type is one of:

I1: Sample index read
I2: Sample index read
R1: Read 1
R2: Read 2

Due to suboptimal results from the cellranger pipeline, I decided to do the reads alignment 'manually'. I plan to use bowtie2 or bwa, but I don't know how to process and use this set of fastq files at the same time. The command is as follows, usually sample.fastq.gz can only use one or two fastq files.

bowtie2 -x mm39 -p 60 -t -U sample.fastq.gz -S sample1.sam

What should I do?

Answer 1

[ATpoint]

Don't. This is far! too complex to be done by hand. You do not only have to align reads, but also identify and deduplicate UMIs and cellular barcodes, assign them to each cell, solve / discriminate between empty droplets and true cells. Also, neither bwa nor bowtie2 are splice-aware, also not aware of the R1/R2 structure of 10X data in terms of connecting cells/CBs to reads. If CellRanger, which is extremely common and well-established indicates a problem then rather try to figure out what the problem is and whether it is related to library prep. Custom approaches won't work. It is an entire project at its own to build single-cell pipelines starting from reads, that is why expert software exists. If you really want independent validation then try alternative quantification pipelines such as https://salmon.readthedocs.io/en/latest/alevin.html, kallisto-bustools or STARsolo.