Respected sir, I have one question regarding relative gene expression level in Allen brain Atlas database.
Sir, if we compare expression level of a particular gene with a particular probe in two brain areas such as
central gray substance of midbrain (CGMB) Expression - z-score: 3.80587, log2 intensity: 5.74661
And Substantia Nigra(SN) Expression - z-score: 0.899768, log2 intensity: 3.49319
Sir, as it is observed CGMB has higher level of expression than SN for that particular gene,
Medam, from above data, can we predict or say how much higher the expression is in CGMB ,than SN for that gene, such as 2 times, 3 times , 4times … Higher?
Thank you very very much sir
Yours sincerely
Pritesh Lodh.
You can predict the difference by subtracting the log2 intensities. log2(fold change in CGMB vs. SN) = central gray substance of midbrain (CGMB) Expression log2 intensity (5.74661) - Substantia Nigra(SN) Expression - log2 intensity (3.49319) = 2.25342m, and therefore it is 2^2.25342 = 4.77 times higher in CGMB. It’s worth noting that both ISH (for the mouse brain atlas) and microarry (for the human brain atlas) aren’t perfectly quantitative, so this should be considered as an estimated fold change.
You cannot necessarily predict protein expression from gene expression, although higher gene expression often means higher protein expression. Studies have found correlations between gene and proteins in the R=0.4-0.6 range (varying widely depending on details of the experiment, but the point is that it’s much better than chance but very much not perfect).
1 Like
Thank you very much Sir.
Sir, as it has been found that 4.77 times higher gene expression in CGMB than SN for this gene.
Sir, if we consider that the expression of that gene in SN is 1 TPM, then can we say the expression of this gene in CGMB is 4.77TPM, as it is 4.77 times higher?
Thank you sir.
I’m not exactly sure since the the data comes from different experiments. I think it would be reasonable to say that based on results from the Allen Human Brain Atlas you expect a TPM value of 4.77 (or something to that effect). I don’t think you can explicitly claim that since you didn’t make the measurement. Maybe others with statistical backgrounds would like to weigh in?
1 Like
Hello Sir.
Sir, if with one particular probe of a gene , the expression intensity in prefrontal cortex is 9.3 and in amygdala is 9.8 ,
Sir, from here can we conclude that, the expression level of this gene , in amygdala and prefrontal cortex is almost same?
Thank you sir
Yours sincerely
Pritesh Lodh.
I think that is a reasonable conclusion–although it is possible that the difference is statistically significant, I doubt it.
1 Like
Thank you sir.
Sir, does it mean, depending upon the probe, this difference may be significant or insignificant?
Your statement is true, although that is not what I meant. I meant that you’d need to perform a specific statistical test (e.g., T-test or related) to determine significance, although the difference between 9.3 and 9.8 isn’t large regardless. If you are interested in deciding which probe to use for a specific gene, see this post.
1 Like
Thank you again sir.
Sir, as it mentioned that probes with lowest q-value best represent true gene expression values as measured with RNA-seq…
Sir, in the column PASS, what is False indication mean for a particular probe for a gene mean? Does the probe with false notation not represent true gene expression and only the probe with TRUE indication is matchable with rnaseq data?
Thank you sir.
Correct. In additional file 8 of our paper, FALSE means that there was not a significant correlation between expression of the same samples based on microarray vs. RNA-seq. This likely means that the probe is inaccurate and/or the probe has very low expression.
1 Like
Sir, I searched for a gene srd5a2, which shows with both probes False notation in PASS coloum.
So it means the microarray is not correlate with rnaseq data.
Sir, how can I find true values of this gene in Allen brain atlas database.?
Sir, is there any relation between TPM value in rnaseq and log 2 intensity value in microarray mathematically?
Thank you so much sir for kind consideration.
Log 2 intensity values in microarray are measures of how bright the spot is on the microarray, which is reliable for differential expression (e.g., twice the intensity typically equals twice the expression), but there is no perfect correspondence with TPM. As part of the same table you are looking at, we share an approximate conversion factor for the subset of probes with good agreement between RNA-seq and microarray. I don’t have a good answer for your question about “true values” of a gene, since every value is based on a measurement.
1 Like
Thank you sir.
Sir, as I search for the gene expression for srd5a2 gene, what is the approximate conversion factor for the probes of srd5a2 gene, between RNA-seq and microarray data?
Thank you again sir.
See columns C&D in the table (y=mx+b) for the conversion. These probes have low expression and did not match between microarray and RNA-seq, so I would advise against using the conversions for this gene. SRD5A2 also has low expression across prenatal and postnatal development, based both on microarray and RNA-seq in the BrainSpan atlas (here and here), so I’m guessing it really does have low expression (or is expressed in a rare cell type).
1 Like