e-mail+to+Kathy+about+analysis+3-7-12

When I just look at the p values from a test rather than looking at the top binders or the fold change, everything works out really well. For example, when I just look at the p value from a t test comparing the SMC1fs sera samples to the naive sera samples, the SMC1fs spot on the array has a very significant p value (0.007851). On the other hand, the p value of the PUC19 spot is not significant at all (0.452377) so this is exactly as we would expect. Additionally, when looking at the top 10 spots with the most significant p values, SMC1fs is in this top 10 list and PUC19 is not.

When I do the t test comparing tumor sera results to naive sera results (rather than SMC1fs to naive as in the 1st paragraph), I can get the top 10 spots with the most significant p values. SMC1fs and PUC19 are not in the list of top 10 or even close to the top of the list for all the spots just as I would expect it should be. Note that 9 out of 10 of these tumor spots decrease from naive to tumor. I've picked the top 5 decreasing and top 5 increasing for further screening (listed in attached powerpoint).

So if I just look at the p values everything works out very well.

Also, I plan to run more slides for the Torino samples. In this last experiment, I had one slide for "transgenic no tumor", "tumor", and "wild type" without any duplicates. However, my wild type slide broke. I can still scan it in our scanner, but our scanner is not sensitive enough for these types of slides. The scanner in LeBaer's lab that I used for the other SMC1fs, tumor, and naive slides works great, but I can't scan a broken slide in that scanner type like I can in the CIM scanner. Therefore, I have no torino wild type to compare to, and no replicates to be able to do t tests and get p values. I plan to print more slides, and run the Torino samples again. I will print 3 slides for "transgenic no tumor", 3 slides for "tumor", and 3 slides for "wild type". 3 slides will be nice because then I'll have replicates even if one slide breaks.

We will push forward with the colony hybridizations for the 4T1 tumor information too though, and won't wait until we have the results from the Torino samples.