Data-driven Science Done Wrong
•Must have some hypothesis – data is not the end goal of science
•Finding patterns in the data is where analysis starts, not ends
•Must understand the limits of high-throughput technology (e.g. microarrays measure transcription only, one genome does not tell you about species variation, etc.)
•Must understand or explore the limits of your algorithm
•
Controls & stats – if you want to see what genes are changing on a microarray, you should have some controls – those you expect should change and those that shouldn’t