# Basic question about finding mRNA sequence in transcriptome

I am a computer scientist just starting out in bioinformatics topics, and would appreciate any guidance that can be given here:

I have an mRNA sequence- an isoform - whose length is about 4000 base pairs (bp) - the origin is provided in the link.

I also have a bam file/fasta file of the entire transcriptome of a kidney cell (see Data Access tab).

My objective is to quantify mRNA expression levels of the mRNA sequence in the transcriptome.

How does one get started? Should I use the original format (TenX bam, about 10 GB) or SRA archive data (fasta, about 3.5 GB)? Why are the files so different in size?

I have a MacBook with samtools installed and 16GB of RAM. Is this enough to process the data and what basic commands should be used to quantify the 4000 bp sequence overlaps in the transcriptome?

• Hi @the_darkside this isn't what we do. The package you need IMHO is called Seurat, it would be recommneded that you read the manual and return with specific technical questions.
– M__
Jul 26 '20 at 7:31
• Right. Looking up gene counts for a single gene in a single sample, it's pretty useless. Especially on a single cell experiment, which is complicated. Is this really a class assignment? Jul 26 '20 at 16:49
• I know that looking at a single gene in a single cell sample is not very useful in and of itself, but the idea is to get some practice in mRNA quantification. I hope that once I understand the process, I can repeat the process on tissue samples. I haven't yet been able to find an available file of non-single cell mRNA sequencing of a human kidney. Jul 26 '20 at 22:32
• This is not how you do it. Find a bulk RNASeq sample, and go through a tutorial. Jul 27 '20 at 0:04

Basically you want to know the VHL Gene expression in that 10x dataset. They have provided the barcode, matrix and feature files on GEO GSE131685 as Supplementary file. You could use those files and get the gene expression matrix as described on 10x website.

library(Matrix)

matrix_dir = "/opt/sample345/outs/filtered_feature_bc_matrix/"
barcode.path <- paste0(matrix_dir, "barcodes.tsv.gz")
features.path <- paste0(matrix_dir, "features.tsv.gz")
matrix.path <- paste0(matrix_dir, "matrix.mtx.gz")

stringsAsFactors = FALSE)

stringsAsFactors = FALSE)

colnames(mat) = barcode.names$$V1 rownames(mat) = feature.names$$V1


This mat can be further loaded into Seurat for any downstream analysis. All this can be done on a laptop.

If you want to re-quantify the gene expression for whatever the reason, you need to re-run the cellRanger pipeline using GTF of your interest (it needs some preparation), which may not work on your laptop. It requires large memory and long running times, ideally to submit as job on cluster.

Should I use the original format (TenX bam, about 10 GB) or SRA archive data (fasta, about 3.5 GB)? Why are the files so different in size?

I strongly recommend you hold your horses; how do you think you can analyze something if you don't even know what you are looking at?

Since this is a "TenX" bam, you can probably do the job with samtools and grep. Looking up 10XGenomics bam tags might be helpful. You might need to look up SAM flags too.

My objective is to quantify mRNA expression levels of the mRNA sequence in the transcriptome.

You understand that this isn't really a thing that's possible? With spike-ins, maybe, but you don't have that. All you can do is work out the proportions of reads belonging to that transcript (Or gene, really, unless you want to put in a great deal more work, beyond I believe what 10XGenomics does). Which I guess you want to do by lumping all the cells together, and just pretending there is no QC to do?