Please create code for this:

Read in a table "data" using the readtable function. It is a 10000 x 39 table. The first column is all the gene names. Each column after belongs to a patient. So there are 38 patients, for which 10000 gene expression values have been recorded.

First z-score your gene expression data by gene across patients. For this, you need to transpose the data. This is your normalized data.

Then, select a subset of the top 5% most variable genes for modeling. Use the un-normalized data and the var, prctile, and find functions for this.

As such:

V = var(un-normalized data);

P = prctile(V,95);

idx = find(V > P); % gives row numbers for top 5% of genes

Take the row numbers given by the idx to get the correct top 5% genes from the normalized data. Create a table "top5Percent" that holds the gene names for the top 5% and the normalized data for each gene.