GV Exercise.1
Find Reference genes from a given context
[ Main_Page | Genevestigator_training | Analyze_public_microarray_data_using_Genevestigator | GV Exercise.2 ]
last edit: October 31, 2014
AIM: Identify candidate reference genes for normalization of qPCR experiments based on a low variance of expression levels across public experiment(s) of biological conditions that are similar to the conditions that you study.
This approach is the first step in selecting reference genes for qPCR and is used in conjunction to qbase+: select 8 to 10 candidate reference genes in Genevestigator, measure their expression by qPCR on a representative set of your samples and select the candidates that are the most stable in your samples using qbase+.
As always in research, you SHOULD do a preliminary qPCR experiment to check your reference genes before spending a lot of time and money into a full-scale qPCR experiment
We first take an example for which answers are provided then you can repeat the experiment with your own genes of interest. If you do not have genes in mind, tools like QuickGO, (TAIR - for plants), or BIOMART can provide you lists of genes related to a given ontology term.
Introduction
Ideally, reference genes have to fulfill at least two conditions:
- they must have a stable level of expression across all conditions being compared (ie a narrow boxplot).
- their overall expression level is preferably similar to that of the gene(s) of interest that are measured by qPCR (there are several practical and theoretical reasons for this).
We have a gene list
Suppose we are working with samples from Arabidopsis thaliana and we want to select candidate reference genes for qPCR.
step#1: select organism and MA platform
As first approach, we take the whole sample collection for the ATH1 22k array (~10k samples). The genes on this microarray form the set of genes that we want to select reference genes from.
step#2: create a probe list from a defined list of genes
We take all TAIR metacaspase gene IDs. These metacaspase genes are our genes of interest in the qPCR experiment. So we want to select stably expressed genes from Arabidopsis with similar expression levels as these metacaspases .
AT1G02170 AT5G64240 AT1G79310 AT1G79320 AT1G79330 AT5G04200 AT4G25110 AT1G79340 AT1G16420
Build a gene selection with these probes
The obtained gene selection looks like this
step#3: run the tool and look for invariant probes in ALL Ath samples that have expression ranges comparable to these of the metacaspases
Setup the RefGenes search
Run the tool and get candidate reference genes
What is mostly important here is that the candidate reference genes show a very similar expression distribution as the metacaspases. This will make them ideally suited to normalize metacaspase expression
Create a new gene selection with 20 found candidate reference genes and call it metacaspase_references
A clear downside of our approach is that we took all samples from Arabidopsis thaliana and we do not have any specificity towards anatomy or conditions. It is recommended to filter for samples that come from the same tissue as the samples that you are studying. So when you take samples from Arabidopsis thaliana shoots you should select the microarray experiments in shoot-apex (259) and repeat the preceeding operation
results for shoot-apex
What do you see special about the selected candidates in either run?
We will see more examples of the effect of choosing the right context in the next exercise.
Create a new gene selection with 20 found candidate reference genes for shoot-apex and call it metacaspase_references2
Identify any perturbation where the metacaspase_references genes show more than 1,5 fold change differential variation using the Condition perturbations tool
Expand all conditions and filter the long heatmap for at least 1.5log differential expression (2.82x) in absolute value
As seen here, the candidate reference genes show differential expression in only 6 sample conditions (3 groups). This is nice as we want to use such genes as reference genes in qPCR and they should not show variability in too many experimental conditions published by others.
From here on, users should choose the 8-10 candidate genes that show the least color change across particular perturbations, test these candidates in a validation qPCR on their own samples and select the most stable ones from the qPCR results.
results for the shoot-apex control probes in the Ath.all context
Download the full exercise workspace
Try it by yourself before expanding on the right!
References:
[ Main_Page | Genevestigator_training | Analyze_public_microarray_data_using_Genevestigator | GV Exercise.2 ]
