Katrina Sherbina: Week 1
The R statistical package was used to read nonnormalized data in the form of a GPR file for all of the microarray chips prepared in the lab. It was found that the script used yesterday to read the nonnormalized data and then to normalize the data was incorrect. Without clearly specifying whether the mean or median values for each labelling dye should be used, R would with some parts of the script use mean values and with others the median values. In going back to the data today, we specified that we wanted the median values to be used in calculating log ratios. Specifically, we used R to calculate the log base 2 ratio of red to green dye was calculated for the Dahlquist wildtype flask 3 t15/t0 microarray chip . The ratio was calculated by subtracting the median value for the red background from the median value of the red foreground and then this difference was divided by the difference between the median value for the green foreground and the median value for the green background. The negative log ratio was calculated because the chip being analyzed was dye swapped. The first five log ratios calculated by R (two of which were duplicates of two of the three genes looked at) were compared to the first five log ratios in the GPR file corresponding to the chip being analyzed. It was observed that the log ratios of R were the negative values of the log ratios in the GPR file which was expected because of the dye swap done on the chip. However, we should ask Dr. Dalhquist why she used a red to green ratio instead of a green to red ratio for the dye swaps. Also, the log ratios generated by R were then compared to the log ratios within an Excel file of the log ratios for all of the chips. There were no duplicate genes in the Excel file. Therefore, an average of the log ratios calculated in R for each of the two duplicated genes was found and compared to the log ratios within the Excel file. It was observed that the former log ratios did not correspond to the log ratios for the three genes on the Dahlquist wildtype flask 3 t15/t0 chip. We need to talk to Dr. Dahlquist regarding how she did her calculations to maybe shed some light on why the log ratios calculated by R did not match the log ratios that were provided in Excel.
Katrina Sherbina 16:38, 20 May 2011 (EDT)