Daniela Cipolletta
Intro
This is new data that she hasn't looked at before, part of her current research project
Comparing t-cells from different tissue (adipose tissue)
Comparing in a different mouse strain (from earlier experiments in this project
Analysis
- loading data into multiplot
- opening a blank ppt presentation
- names the slide
- waiting on upload
- starts setting up axes of a graph (on ppt), makes two graph setups, draws more labels
- getting her slide set up while she waits on upload
- has a side-by-side multiplot view
- loads up each graph
- turns graphs gray (familiar with graph design?)
- selects some outliers, looks at sidebar table
- changes graph 1, selects outliers again
- saves the result
- seems frustrated?
- generating and saving graphs frequently
? she's highlighting with a previously discovered signature that she doesn't expect to be present here
- adding lots of highlights
- propagates highlights to view 2
- looking at outliers
- removing genes based on some criteria
- highlighting several specific genes
? highlighting the transcription factors - hallmark genes of this (?)
variance between knockout and wild type mice
- generates a volcano plot (p-values)
- propagates highlight, looks intently back and forth between the two
- filters
- looks at significant genes
keeps needing to resize the window to change the settings and get back to a big view
doesn't always use the multiple views in a linked fashion
- nods to herself
? comparing FAT vs. something else
wild type model significantly more represented
strange bias toward wild type
signatures that are either over-represented (red) or under-represented (blue)
daniela reaches conclusion: knockout can move signature in this direction
new information, but in line with previous observations
- nods again
looking at where stuff is... fitting expectations, etc.
- propagates highlight again
mostly seems to be doing manual linking
- goes to get a colleague for discussion
(videos break, sound probably goes slightly off-sync...)
- returns
- clears some highlights
looking for a signature given by jaime. from another platform, so it needs to be translated.
- finally finds the signature and puts it in her graphs
- has trouble seeing what highlights are on due to screen space issues
- puts highlight in view 2 and looks back and forth.
Post-analysis discussion
mice with transcription factor
fusion with protein allowed genes to behave properly.
looking if cells with these hallmark factors are different
checked first replicate a month ago
this is a follow-up experiment
process steps:
what is the variability?
filter for most solid data
looking from genes to see how they change
previously, some genes don't belong to this, appearing in knockout mice
check if there are still weird genes and find them
next, to make a gene list of differnt experiments between animal models, study to see if they follow in a particular pattern.
explain phenotype in disease?
likes to have two plots to see two different comparisions for same population
see significance as well as behavior
if I don't have 2 I have to go back and forth
4 is even better
sometimes you do multiple comparison - many populations
going back and forth you can forget and lose time
big files, 100 samples, 30 populations
wants to grab multiple highlight files