Saturday, February 04, 2012
Catch Us If You Can
This image struck me as very interesting. I pulled it from the Abnormal Science blog. I've blogged about my own experience with the same assay. The concept is simple. You culture RAW 264.7 cells in the presence of RANK-L. In about 5 days this group of cells will fuse together resulting in a group of ugly multinucleated osteoclasts. By adding an antibody against the RANK-L or it's receptor, you can retard or prevent osteoclastogenesis. You could also try RNAi against a critical protein in the pathway kicked off by the RANK-L. Either way, that office bound scientist needs the picture we see above.
My experience with this assay was with RNAi. The very first time I used RNAi I obtained the results that we hoped for. I could not however, repeat those results. I still don't know what happened. I used RNAi against IL-2 which was accused of being in the pathway by a bio-informatics program we were selling. This evidence showed that we were geniuses. In spite of the 9 times this assay did not work, we had photographic evidence that it did. No one was going to ask us how many times we tried this and what was the outcome.
It is not at all difficult to imagine what your P.I. wants when he designs an experiment. Especially if your P.I. is in the business of churning out publications. If he/she says, "provide evidence of osteoclastogenesis and the inhibition of it with my awesome idea", then that is what will be accepted. If you can't do it, someone else will. Do you want to be important and successful like the P.I. or do you want to argue with him/her. P.I.s don't like dissention.
In my example, I obtained the desired outcome right out of the gate. After a few more failed attempts the assay was passed on to another person who gave a presentation of his work on the assay. It was 'sort of working", he claimed. The TRAP assay, which measures tartate resistant alkalyne phophatase, can be done but on a separate petri dish of cells. So for each condition two groups of cells are treated in order to obtain visual and enzyme assay evidence of osteoclastogenesis. If you want to test a treatment more than once, as good science would require, you would need two groups of cells for each replication. What the new lab researcher did was to run the TRAP assay on the same group of cells being photographed at day 5. TRAP accumulates on days 3 and 4. It is the same at days 1,2, and 5. There is a flaw in experimental design. But the P.I. is looking for good images for a powerpoint presentation. He isn't receiving a presentation on experimental design.
Areas on the petri dish where the cells were not fusing were captured in the knock-out treatment. Areas where the cells were fusing in the RANK-L only control were captured. Without using photoshop, the desired images were obtained. The P.I. was happy but the images were not as good as my originals. The researcher had to keep trying.
The new researcher could not hide the fact that the enzyme assay quantified the enzyme being produced for the entire plate of cells. He could not cut out a small section of the plate as he did with his camera, and measure TRAP. What does one do? When showing data on a powerpoint you have a biogolical science tool that has been used over and over. Run the assay over and over. Find the measurements, both within a margin of error, where the measurements differ the most in the direction required by your P.I. Make a bar graph from the data and stretch that graph out to make the height of the bars tell the story you are working on.
When I saw the obvious re-use of the image of the Zerumbone treated cells at days zero and 3, I thought perhaps they could be two different pictures of the same plate. I would have no way of knowing what day they were taken. The darkened circular outline on the right side of the image gave it away. The filter inside the imager created that outline. With a little more effort they could have slid the filter out of the way and photographed a different section of the dish. Dishonest yes but at least not an insulting lack of effort. The re-use of the same picture shows how easy it is to use photography to deceive. You do not even need to use photoshop.
In the concept I put forth in the last post, a Consumer Report for Science Consumers, a small group of image experts would offer non-biased opinions on the images used in each publication being analyzed. The area of expertise would not be in osteoclastogenesis.