How to remove non protein coding genes from single cell pre-mRNA seq?

Question

[fifty_fifty]

I have an output from cellranger with gene expression for each cell (pre-mRNA seq). I need to remove all non-coding gene from there. What is the best way to do it?

[caggtaagtat]

What is pre-mRNA seq? Do you mean total RNA seq?

[fifty_fifty]

pre-mRNA is a primary transcript that contains both introns and exons

[caggtaagtat]

I know, but how can you only sequence pre-mRNA? Capping and poly-adenylation is happening after transcription, but splicing is already happening during transcription, so I'm genuinely curious, because I never heard of pre-mRNA seq.

Answer 1

[caggtaagtat]

If you want to subset the bam file you can do:

samtools view -h -b -L genes_coordinates.bed in.bam > out.bam

where genes_coordinates.bed is a bed file with the genomic coordinates of each coding gene. This information could for example be downloaded from ensembl (biomart).

[ATpoint]

Does that respect overlapping genes on different strands?

[caggtaagtat]

This would be independent of strand, so there would still be some reads left which do not belong to mRNA, but e.g. lnRNA transcribed from the other strand.

[ATpoint]

I would simply make the count matrix, get a GTF file and keep only what is annotated as protein_coding.

[fifty_fifty]

How to get a GTF file from cellranger output?

[ATpoint]

You have to download it from any of the standard repositories, e.g. GENCODE, NCBI. it is a reference annotation file. Do you have a background in NGS analysis? No offense but if you are stuck with removing some genes than the single-cell analysis will be..."fun". But seriously, you should spend some time with the basics.

[fifty_fifty]

I do not really have a biological background. But I am learning. So, I thought I could ask questions here to get a better understanding of things like that. Thank you for your suggestion though

[caggtaagtat]

Yeah, when the goal is a count matrix, that makes more sense.

[fifty_fifty]

thank you. Is there any way to use .h5 file instead?