For scientists

Field guides for bench scientists

Powerful computational tools are now one download away, and so are their traps. These guides are for people who run experiments for a living and want answers they can trust.

Each guide is built around the mistakes I have watched good scientists make, myself included, and around one rule that I now apply to everything: a negative result only means something when a positive control has gone through exactly the same analysis.

They are not programming courses. They assume you can run a tool, or can ask someone or something to run it for you. They are about what to ask, what to check, and when not to believe the answer.

• AlphaFold for experimentalists

  How to get real answers about proteins and complexes from predicted structures, and how to tell when a prediction is fooling you.

• Public RNA-seq and microarray data: reanalysis and GSEA done properly

  Thousands of datasets are free to reuse. How to check that the samples are what the labels say, and how to read a pathway result without over-reading it.

• Pooled CRISPR screens: from read counts to hits you can trust

  Screens are expensive. What a hit is, what a bottleneck does to your statistics, and how to choose what to validate.