We are utilizing a new technology in our lab. The GeXP, among other things, provides a resource for the confirmation of MicroArray data.
Basically, microarrays measure the level of expression for tens of thousands of genes on a single slide for a given sample. In many cases, a small sample of this data that represent a significant differential expression between a sample and a control will be confirmed through PCR (polymerase chain reactions). The drawback of PCR is that only one gene can be analyzed in a single well on a plate. However, the GeXP provides a way to use gene multiplexing to measure gene expression in a similar way but for more genes per well.
I went through some extensive training on this technology and we completed our first experiment (after much trial and error) and I completed what I considered a relatively simple analysis comparing the results from the GeXP with the results of the MicroArray. I simply confirmed that there was a significant linear correlation between the two technologies using the standard test statistic based on Pearson’s correlation coefficient.
I sent a question to my trainer from Beckman Coulter to ask for advice on what type of analysis should actually be done to compare the technologies. I was careful not to tell him what I did because I wanted his independent opinion on the analysis.
Do you want to know what his suggestion was?
"I think you should use bar graphs."
Bar graphs? That’s the deep analysis I need. Of course, how could I not consider drawing pretty pictures as a way of demonstrating the reliability of these technologies. Bar graphs are my dream come true.
I honestly laughed out loud in learning the depth of thought given to the data analysis by the folks at Beckman Coulter. I’m waiting to hear if his team (supposedly of technical analysts) might provide deeper insight. I’m still waiting.
I don’t mean to sound too critical but if this had been a question on a test in my Elementary Statistics courses, it would definitely not pass muster with me. It just shows a lack of understanding of the importance of statistics in data analysis.