7
$\begingroup$

This is probably a stupid question because I don't fully understand how microarray experiments work. I'm trying to understand how microarrays work but I'm confused about something which I can't seem to get around without fully understanding how microarrays work, which seems impossible without understanding this.

I understand the principle of how microarrays work. But I don't get how microarrays can tell you how much mRNA is expressed in one kind of tissue vs another kind of tissue. Wouldn't that depend on how much tissue you started with? So if you started with more cells of one kind you will get more mRNA/cDNA and so the intensity of fluorescence at the corresponding spot on the microarray will be greater. So even if the amount of gene expression is lower in one type of tissue the corresponding spot on the microarray might show a greater fluorescence intensity compared to the other type of tissue if more tissue was used.

Also another related question: if you put the mRNA you obtain from cells through PCR will you get the same proportion of mRNA as before PCR? From what I understand PCR can be used to make more mRNA before doing the microarray but because PCR is a random process I'm not sure if it will affect the proportion of mRNA.

Hopefully my questions make sense.

$\endgroup$
5
  • 1
    $\begingroup$ While someone is no doubt putting together an excellent answer, for some good visuals, Google "Microarray presentation" and the first few hits are good powerpoints for microarrays. $\endgroup$
    – CKM
    Feb 21, 2016 at 2:12
  • $\begingroup$ Briefly, you can compare different samples, which may have started out as different amounts of tissue, by using control genes. The numbers reported by most standard microarray experiments are not absolutely quantitative (There are 2450 copies of this RNA per cell in this sample), but are rather a proportion as compared to a set of housekeeping genes whose levels are not expected to vary much between the samples you are analyzing. $\endgroup$
    – MattDMo
    Feb 21, 2016 at 2:27
  • $\begingroup$ What about PCRs? do you get the same proportion of mRNA after applying PCR? I would still like an answer to the second question $\endgroup$
    – liyuan
    Feb 21, 2016 at 9:52
  • $\begingroup$ @MattDMo gave a great answer. For PCR amplification, you are right, this is a random process and while you would expect most RNAs to be similarly amplified, biases can and will occur. Amplification biases are usually assumed to be identical between similar samples and therefore will (mostly) cancel out during the differential analysis. $\endgroup$ Feb 21, 2016 at 17:49
  • 1
    $\begingroup$ Housekeeping genes are rarely used in microarray analyses, and there are many issues with finding housekeeping genes by itself (See ncbi.nlm.nih.gov/pmc/articles/PMC4430495 ) . Models that work with microarrays usually work with making distributions between samples comparable based on the notion that most genes will not be differentially expressed between two conditions. This can involve scaling, or loess normalisation, or the very popular quantile normalisation. RNA-seq analyses oft use library size or upper quartile normalisation to ensure comparability. $\endgroup$ Mar 10, 2017 at 3:24

1 Answer 1

1
$\begingroup$

From the comments:

MattDMo:

You can compare different samples, which may have started out as different amounts of tissue, by using control genes. The numbers reported by most standard microarray experiments are not absolutely quantitative (There are 2450 copies of this RNA per cell in this sample), but are rather a proportion as compared to a set of housekeeping genes whose levels are not expected to vary much between the samples you are analyzing.

cagliari2005:

For PCR amplification, you are right, this is a random process and while you would expect most RNAs to be similarly amplified, biases can and will occur. Amplification biases are usually assumed to be identical between similar samples and therefore will (mostly) cancel out during the differential analysis.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .