Interpretation of Salmon quat.sf result

Question

[saadleeshehreen]

Hi,

I trying to quantify whether my gene of interest is expressed in a particular bacterial genome or not. From the quant.sf file of salmon, I got TPM 1000000.0 and NumReads 111.00 for that particular gene.

What does this mean? Is the value of NumReads low to conclude that the gene is expressed?

Thanks in advance.

[Rob]

Looks like you mapped only against that gene? A TPM of 1000000 means all of the relative expression is allotted to that gene. This seems very unlikely if you've quantified the sample correctly (using the whole transcriptome as the reference).

[saadleeshehreen]

Hi,

I downloaded the SRA file of the corresponding genome and used the nucleotide sequence of my interested gene. My interested gene is 300 bp long. As I am going to check its function, my intention was to see whether that gene expressed in that particular genome or not. If not expressed, I will not proceed for downstream experiments.

I think it is a clear indication that the gene was expressed in that particular genome, isn't it?

Answer 1

[h.mon]

I think it is a clear indication that the gene was expressed in that particular genome, isn't it?

No, I disagree. Strictly speaking the way you quantified is incorrect and does not allow your conclusion. As Rob pointed out, you didn't use Salmon correctly, and the way you used it has two potential problems:

1) if you map to only one gene, reads that would otherwise map perfectly to other (similar in sequence, but different) genes and not map to this gene, may now may map to this gene;

2) if this gene has high similarity with other genes, reads would map to multiple locations, but their quantification could be ascertained accurately due to Salmon EM algorithm. Now, in the absence of these similar genes, you may be over-estimating the counts.

Use Salmon as intended: quantify the reads against the whole set of transcripts from your species, then examine the counts and TPM for th gene of interest.

[saadleeshehreen]

Thanks for correcting me. I have downloaded the cDNA of the genome from Ensemble and examines the TPM and Numreads again.

The TPM is: 299.9976 and Numreads: 87 Is it now a clear indication that the gene was expressed in that particular genome?

Cheers

[h.mon]

Yes, it is a clear indication it is expressed. You could plot a histogram of TPM values to have a visual indication of the level of expression of your gene of interest compared to the rest of the genes.