Skip to content

Week 5

Jump to bottom
Jasmine Dumas edited this page Jun 28, 2015 · 5 revisions

Here are my tasks for week 5

Task: Develop survival analysis modules including generation of Kaplan-Meier Curves and allow user selection of survival measure (remission, death, end of study, etc)

Method: Research and apply the Cox-Proportional Hazard models; Utilize the survival R package to create routines linked with extracting specific genes and paired with phenotypic data by utilizing the survfit() function.

June 23

  • I'm still trying to understand and get the editable tables to work. Dr. Dancik has been immensely helpful in guiding me towards comprehension and some working code but I think that my solutions are close since the console is printing out the actual replacement values but the new table values.edit$table is not be displayed.

  • The goal for this week was to start and get a working model of the Survival Analysis and obviously getting the tables to be edited is the only road block right now.

  • I'm researching some shiny resources to help my understanding of reactiveValues() and this tutorial from RStudio is making the most sense right now.

June 25

  • Editable tables are now working! in addition to that the table is re-rendered to the original value if the initial GEO submit button is pressed.

  • I have also implemented a gsub() component to allow for partial find/replacement. There really is a difference in x[] and x[[]] and also reading the documentation saves time and boosts comprehension 👍 So the user will be able to simply select a checkbox for either partial find/replace within a cell or unselect it and do complete find/replace.

  • I have a nifty survival example script from Dr. Dancik that includes a plotting function and have laid the groundwork in the form of comments and a new bsModal window on the Survival tab that will allow the user to select the appropriate (pre-edited) columns to include in the function.

  • I still need to finish writing a function to return a sample of the boxplots for the Expression profiles instead of all of them (ie type in GSE13507 and witness true cluttering of all 200+ samples!) & I need to include a way to handle special characters !@#$%^&*()

June 28

  • Several things were updated/fixed:

  1. The auto-detect for the log2 transformation was updated from a or (|) statement to an (&) and statement

  2. The title of the expression plot was updated to reflect either "selected samples" or "samples" depending on how many samples are returned.

  3. There is now a method to return only 30 samples if the number of samples in the data set exceed that which improves the graphics

  4. The full clinical data table had its overall filter removed and the summary table had all of its filters removed

  5. My mentor Dr. Dancik made some awesome changes to the UI which were merged from a pull request!

Clone this wiki locally