Regarding FDR P-value calculation in JMP

Question

Hi,

I have performed two sample nonparametric test (Wilcoxon) to see the significance difference between male and female for approx. 100 gene gene expression in JMP. Now, I would like to see FDR p vale for this analysis. Could you please suggest me that how can I  perform this analysis in JMP?

Thanks

Neeraj

txnelson · Answer

I dug up an old piece of JSL that I used to calculate the FDR.  Attached is an example data table that is the results from a Response Screening platform.  I cut it down to have just the PValues and the FDR values calculated from the Response Screening platform.  The script below, takes the p Values from the data table and independently calculates the FDR.

Names Default To Here( 1 );

dt = Current Data Table();

pValues = dt:PValue << get values;

index = Rank Index( pvalues );	// removes missing values
n = N Row( index );

ordPValue = pvalues[index];
adkPValue = J( n, 1, . );

For( i = 1, i <= n, i++,
	k = n - i + 1;
	ratio = n / (n - i + 1);
	adkPValue[k] = If( i == 1,
		ordPValue[k],
		Min( ratio * ordPValue[k], adkPValue[k + 1] )
	);
);
dt << New Column( "kEN Calculated FDR PValue", Format( "Scientific" ) );
For( i = 1, i <= N Rows( adkPValue ), i++,
	dt:kEN Calculated FDR PValue[index[i]] = adkPValue[i]
);

example.jmp

Neeraj · Answer

Thanks!

Thierry_S · Answer

Hi Neeraj,

With a bit of digging, I was able to find an Add-In developed by John Sall from JMP that calculates correctly the FDR p Value False Discovery Rate PValue

If you are not familiar with Add-Ins, just download the file to your computer and double click in the file: it should install automatically

Best,

TS