Wednesday, February 16, 2011

Data oh Data what are you telling me

Last year I spent a considerable amount of time writing up a mock grant. It had to contain all the thing a real grant would contain, objectives, rationales, specific aims, hypotheses etc. It was a challenge, but won that really helped me to learn more about my subject area and more about what could or could not go wrong. Painful as it was, I did learn something from it. The last few months (since my comps), when I wasn't suffering from nausea, pregnancy induced sleepiness, or crazyness I have been tackling those specific aims. Much to my PI's delight, I have very clear and convincing data. My images are beautiful.  Except the data isn't showing me what I predicted. Yeah, yeah I know cry me river! Hear me out thought

Based on in vitro vertebrate data, I hypothesized I would see a particular phenomenon, in vivo, in my invertebrate model.  (We should all know that I work on the most amazing and wonderful mode organism, the venerable Drosophila melanogaster.) Thing is, I'm clearly not seeing the phenomenon seen in vertebrates.  So WTF is going on!!!

I know I shouldn't complain I should be happy on multiple fronts. The main thing I should be happy about is that  the data is clear and convincing AND more importantly my PI happily accepts clear and convincing data, regardless of whether or not its what she wanted to see.   This is huge, because I have worked for the PI that just had me redesign and redesign the experiment to get the result he wanted. Not. Cool.

Secondly, my treatment is having an effect - I do something to protein X and I get a phenotype. My phenotype is very clear and unique to my protein of interest.  The question is, what is the mechanism of the phenotype? What is the function of X, such that its loss is causing A?  We / I though it was due to a mechanism hypothesized to occur in vertebrates, based on similarity of our proteins and the structures they are involved in. But as I just said that is not happening.  Which means I have to come up with a new theory....which is painful and hard for me. Luckily, I don't have to do it right away. I want to confirm that the phenomenon does not occur by doing some additional controls. These will not take alot of time and will confirm that the phenomenon does not occur vs my assay not being sensitive enough to detect the phenomenon.  No, I'm not beating a dead horse. The mechanism I am looking at is a signalling pathway and I've only looked one step down. There might not be enough of a change to detect, so I'm going to look further down the pathway to be sure that there really is no change.

I shouldn't say I've been in data hell. Since its good, clear data. But something is going on with my Filezilla **such that its downloading files off the server at extremely slow rates (30Kb/s vs its normal >150Kb/s). Considering I have image files near 1G each, its been a frustrating few days. Slow and frustrating because it takes forever for me to get a file, but only about 10 minutes to analyze it. 

But I guess I get lots of time to think about what is going on.....

and PhysioProf, if you're reading this dont freaking tell this is what science is.  Just. don't.

***if anyone knows what could be going on with the Filezilla please let me know.

1 comment:

The bean-mom said...

As you said yourself, SM, what is important is that the data is clear and convincing. It doesn't support the hypothesis that you thought it would... but often that is exactly when science gets interesting.

Something similar is happening with my project. I knocked down a gene and got a certain phenotype. The data was beautiful, and I had ideas about what it meant... and now things are getting complicated, and it's not as straightforward as I thought. But it almost never is, is it? =)