Dalia Burzyn

Intro

New data from someone's experiment

comparing 2 different mice types with t-reg and depleted t-cells

induce a muscle injury to see t-cell effect on muscle regeneration

gene expression in the muscle

not used to analyzing this sample - unfamiliar domain

did some pre-processing that i skipped recording due to low videotape (similar to jaime)

Analysis

just exploring, making comparisons

• generates a plot
• selects outliers, looks at table
• highlighting
• looking a lot at the table

? looking for immune-related genes

hypothesis is that they'll be up-regulated

also need to look up gene names

• looking at some more comparison
• leaves genepattern
• finds some older graphs from an excel sheet
• working with excel

? file in excel is the one used to generate multiplot graphs... excel file with all genes named and with expression value for each sample.

multiplot doesn't show gene names, just probe number

to find something specific, she looks at excel

• looking thoughtfully at excel sheet
• searches for another gene
• startled when she sees a weird shape!
• investigates further by looking at genes in the sidebar table
• replots, replots again... trying several different plots
• looks at the email where she received the data. produces some kind of quality chart, but she doesn't know how to read it.
• finds another weird shape, scans genes in the table.

this file has old and new samples together

some old samples were bad quality

the weird shape is bad data

• wants to split up old and new samples instead of averaging them

.. i run out of tape, there's a gap while i switch it out

more data problems! the + and - in a sample name are getting cut out in a pull-down menu, and that's the most important information in the sample name

• asks daniela for help, they go back and forth
• tries again

at this point, would ask scott or the student who generated the data for help

? (what does bad-quality data mean?)

have tissue or cells, extract RNA, submit it to a company that does the microarray assay, and they send back data

when data is sent outside, who knows what happens to it?

problems can be contamination, broken molecules, RNA destruction

or processing issue at outside company

can mean issues with a whole batch of data