Second exercise on analyzing gene expression data in qbase+

[ Main_Page | Exercises on using qbase+ ]

In this example we will analyze data from another expression study with the following characteristics:

All samples fit in a single run: Run7
We have the following samples:

• 5 control samples: control1, control2…
• 5 treated samples: treated1, treated2…
• 1 no template control: NTC

The expression of the following genes was measured:

• 2 reference genes: refgene1 and refgene2
• 2 genes of interest: gene1 and gene2

There are two technical replicates per reaction

Creating a new experiment

Create a new Experiment called GeneExpression2 in Project1
You can find the details on how to create a new experiment in Creating a project and an experiment

Loading the data

Import Run7. This file is in qBase format.
You can find the details on how to import the data file in the Loading the data into qbase+ section of Analyzing data from a geNorm pilot experiment in qbase+

Adding sample annotation

Download the the sample properties file.

Add a custom sample property called Treatment.
You can find the details on how to add a custom sample property in the Adding annotation to the data section of Loading data into qbase+

Analyzing the data

Choose the type of analysis you want to perform.

Check controls and replicates.

First set the minimum requirements for controls and replicates You see that 6 replicates do not meet these requirements (red). Select to Show details and manually exclude bad replicates All negative controls pass the test (green). Positive controls were not included in this analysis. Qbase+ will now open the results for the failing replicates: as you can see the difference in Cq values between these replicates is not that big. They fail to meet the requirement just slightly.

Which amplification efficiencies strategy are you going to use ?
You don't have data of serial dilutions of representative template to build standard curves so the only choice you have is to use the default amplification efficiency (E = 2) for all the genes.

Appoint the reference genes as reference targets.
You can find the details on how to appoint reference targets in the Normalization section of Analyzing gene expression data in qbase+

Is the stability of the reference genes ok ?
In the Reference target stability window the M and CV values of the reference genes are shown in green so the stability of the reference genes is ok. You can find the details on how to check reference target stability in the Normalization section of Analyzing gene expression data in qbase+

Which scaling strategy are you going to use ?
Since you have a treated and a control group, it seems logical to use the average of the control group for scaling. You can find the details on how to specify the scaling strategy in the Scaling section of Analyzing gene expression data in qbase+

Look at the target bar charts.

In the target bar charts group the samples according to treatment.
You can find the details on how to group the samples in the Visualization of the results section of Analyzing gene expression data in qbase+

The samples of each group are biological replicates so you might want to generate a plot that compares the average expression of the treated samples with the average expression of the untreated samples.

In the target bar charts plot the group averages instead of the individual samples.
In the Grouping section at the bottom of the chart you can select Plot group average:

Are there any genes for which you see a clear difference in expression between the two groups ?

For gene 1, the mean expression levels in the two groups are almost the same and the error bars completely overlap. When you look at the title of the Y-axis, you see that 95% confidence levels are used as error bars. In case of 95% confidence intervakls you can use the following rules: if they do not overlap: you are certain that the difference between the means of the two groups is significant if they do not overlap: you know nothing with certainty: the means can be different or they can be the same So for gene 1 the means are very close but just based on the plot we may not make any conclusions with certainty. For gene 2, the mean expression levels in the two groups are very different and the error bars do not overlap. So the 95% confidence intervals do not overlap meaning that we can be certain that the difference between the means of the two groups is significant. Note that the average expression level of the control group is indeed set to 1 by our scaling strategy

Use a statistical test to compare the expression levels between the two groups of samples ?

You only have 5 replicates per group so you cannot test if the data comes from a normal distribution. Qbase+ will assume they're not normally distributed and perform a non-parametric Mann-Whitney test. The p-value of gene2 is smaller than 0.05 so it has a statistically significant difference in expression levels in treated samples compared to untreated samples. For gene1 the p-value is 1 so we have no evidence to conclude that the expression of gene1 is different in treated compared to untreated samples. You can find the details on how to compare the means of the two groups in the Statistical analysis section of Analyzing gene expression data in qbase+