hello everyone,
First, I will take this opportunity to clarify the course note evaluation. It will be based on a report exactly like that of TP 7 - ie, identifying DE (differentially expressed) genes between 2 conditions, then carrying out a cluster analysis on one of the conditions. The only difference between TP 7 and the final report is that the final report will be on a different data set.
Next, for those of you also taking my course in Applied Biostatistics, you may use this report as your individual report for the biostats course. You may NOT, however, use any biostats topic for both courses. You MUST do a genomics report. If you do decide to do the same report for both classes, you only need to submit it once (to genomics class), and you should email me to let me know that you are doing one report for both classes (if you have not already).
And finally, just to summarize what you need to do in your report:
- short introduction to the study (for TP 7 you are comparing gene expression between placenta and testis)
- Quality assessment - you can start with EDA (e.g. boxplots of fluorescence intensity for each chip), you need to do RMA-QC (minimally weights plots for each chip and NUSE boxplots). You must explain M-estimation to show how you obtain the weights. If you remove any chips from further analysis, justify (i.e. median NUSE > 1.05).
- RMA for the chip set, explain RMA in terms of the 3- step process (bg correction, quantile norm, chip effect estimation)
- Statistical analysis of DE: including your linear model, design matrix (and contrast if necessary); explain the multiple testing problem and what correction you use (BH will be most common but you can use something else if you justify); explain your gene ranking (e.g., by increasing p-value / abs mod t; if you use B-stat you also need to explain the B-stat model); give an estimated number of DE genes and explain how you got this (what is your threshold for declaring a gene 'DE'); any appropriate plots (e.g. relevant volcano plot)
- Cluster analysis: completely explain what algorithm you use, include relevant graphics, interpret clusters (if possible)
- Summarize your conclusions
- Gene list of top 50 DE genes on a separate page (that does not count toward the page maximum)
- (Reproducible) R code - your R code should be included as a plain text file (ASCII, not rtf, not doc, etc), and I should be able to reproduce your output if I execute that file
- I will also evaluate the Overall presentation, including clarity of explanations, appropriate citations / references. Any plots / tables should be 'pretty'.
Questions? Problems? Please don't hesitate to email me...... I will be in my office (MA B1 477) this week Thursday ~11-13, and Fridays 12-13. If those options don't work out for you, please let me know and we can schedule a time to meet, either in person or by zoom.
Best regards,
Darlene
