Hi,

Maybe you could try a correlation test using concentration/density data on each protein from several replicates. It might give you a clue on the strength of association between the two proteins and the direction of the relationship (direct or inverse). But, be careful, you must have certainty that there is a causal relationship between the two proteins.

However, in this sense, even when not much technical details are given in your question, there are two things that troubled me: western blot and difference of expression. As the quantity of transcripts cannot be measured with an SDS-PAGE and less of all with a western blot (which would only show protein(s) recognized by an antibody), it is void to associate the differences observed on the recognized bands as directly caused by differences in protein expression. Thus, to be sure of the causal relationship behind the differences on the bands corresponding to the two recognized proteins, you must re-analyze your results, to carefully consider all possible causes (e.g. different biochemical properties of A and B leading to differences in extraction/transfer or non-saturating quantity of the revealing antibody leading to competition), to do some troubleshooting to discard the different possibilities and see afterwards if a correlation coefficient could fit with your biological issue.

Hope this will help.

Good luck!

Annia

Assuming you've done all the due diligence suggested by Annia Alba above, why not simply plot A and B on the X and Y axes respectively? If both signals are derived from the same blot, there is arguably not even a need to normalise (as you're always measuring a comparison within the same sample, and a strong relationship should hold up regardless). If you have a pseudolinear relationship (such that A+B always approximates to the same total) then this will show up as a straight linear inverse correlation. If you have a pseudolog relationship (such that A*B always approximates to the same total) then this will show up as a linear inverse correlation on a log/log plot. If your expression pattern is such that the switch is always dramatic (cells go from High A, Low B to Low A, High B with little time in-between), then you'll effectively see two discrete populations on your plot.

All of these, I would argue, present the correlation quite clearly to the reader.

Hi

You need to perform linear regression analysis of data of protein A and B after normalization with corresponding housekeeping protein. Simply plotting the data of A and B on X and y axis respectively will give information regarding how strong is the association.

Check the article

"Regulatory factor X1 is a new tumor suppressive transcription factor that acts via direct downregulation of CD44 in glioblastoma"