How do I order my data for analysis in Prism by Mann Whitney U test for nonparametric data ?

I have two conditions to extract proteins in replicates of 3. There are many blank results in the replicates but I am still interested in this data. My main aim is to examine the distribution of certain groups of proteins between methods. The journal I will submit to, ISME requires comparisons to be done by Bland Altman if data is normally distributed. However my data are not normally distributed. Quantification is by NSAF.

I have been trying various options in Prism and receiving different answers, both significant and non-significant .

I can sum the 3 replicates for each protein, make a coulmn and then compare this for the two extraction methods. With or without blank values. This gives me more significant values.

Or

I can enter all the values from one extract matched to corresponding values in the other extract. With blank values I can do a more pairwise comparison but these results rate as less significant . Also in some comparisons the proteins are not detected by one of the extraction methods which makes comparions difficult.

Any advice ?