>>11984901Cell biology, funny enough. My work was 50/50 wetlab and computation during my PhD. Decided I liked data more than cells. Data doesn't just up and die on your for no reason.
>>11984820>datasetgeneric word for gigantic table of data. What big drug companies do a lot of is set up giant, automated assays, where they test every molecule under the sun against some known target of disease (for example, maybe a specific protein is needed for HIV to infect people, and you can look if a compound reduced or gets rid of that protein).
Tens of thousands to millions of data points. It becomes unwieldy after a while. Or the real big one, is people who comb through literature, and extract information from tens of thousands of paper and sums up all the results of, say, known molecules to prevent HIV infecting cells.
>curate?These datasets are usually giant flaming garbage piles because everyone uses 25 different words for the same thing, put values in the wrong columns, forget to say that they were looking at a mutant instead of a normal HIV infection, etc. So if you just took it and built a model with data as-is, you get a shit model.
Curation is the fun term for "going through the data, throwing out the shit, keeping only the good stuff". Its unsexy, boring, and takes 70% of the entire process of machine learning and modeling. Therefore, its a bottleneck and is not being addressed like at all. 100 papers on sexy neural nets, and 0 papers on the 8 months they spent curating data.
GUI - graphic user interface. For dummies who can't code, but want to curate.
