How to find and visualize expression level of a dataset?

Question

[uveyik]

I am dealing with cancer data, VEER in R. This dataset contains 24481 genes and 78 patients. I need to apply a series of data science operation on this dataset in R. Matrix containing gene expressions can be obtained with exprs(dataset).

I loaded the library, first. I need to find out if the overall expression levels of patients similar to each other with respect to standard deviation and the median of expression together with plottings.

a<-exprs(VEER)
medianV <- apply(a, 1, median)
meanV <- apply(a, 1, mean)
head(medianV)
head(meanV)
plot(meanV,medianV)

I obtained the following graph. How can I interpret this graph with respect to the question if the overall expression levels of patients similar to each other with respect to standard deviation and the median of expression? Or how can I find this?

Answer 1

[BioinfGuru]

"How can I interpret this graph?": The graph you created only tells you the the relationship between meanV and medianV.... which is intuitively obvious anyway... in ANY dataset, when mean goes up, so does median. The plot doesn't tell you anything useful about YOUR dataset.

To find "patients similar to each other with respect to standard deviation and the median of expression" means you need to create subgroups in the dataset. For that you need clustering which will create subgroups of patients with similar expression profiles. Take a look at the this to get you on your way. There are many ways, I suggest a heatmap with whatever clustering method is the default in the heatmap R package you use.

Once you have your subgroups, then you can plot the stats of each subgroup with a boxplot.