8.2 Treat BLAST Searches as Scientific Experiments

Scientists are often taught to structure their experiments into four parts: the question (hypothesis), the approach (experimental design), the results (data), and the interpretations (beliefs). This approach shows that beliefs depend on the experiment's data. Whether or not an experiment is capable of answering the question is one way to separate good science from bad.

When setting up a BLAST experiment, the most important thing to remember is "you get what you look for." In other words, search parameters determine what you find. For example, the BLASTN program from NCBI with the default settings assumes that the alignments you are seeking are nearly identical because the parameters (match +1, mismatch -3) have a target frequency of 99 percent identity. If your experimental question is "How many worm genes are related to my favorite human gene," using the default parameters would be foolish because the approach (looking for nearly identical sequences) isn't expected to answer the question; too many sequences have changed in the 500 million years that separate worms and humans.