0%

Bioinformatics as a science

Have you used BLAST? Is a BLAST search an experiment?

Figure 2 Example of BLAST visual output.

Here’s an example hypothesis: this yeast protein is encoding a function that’s conserved across eukaryotes. You could test that by running a BLAST search and see if you find it in other eukaryotes?

Therefore, you may be able to use BLAST to confirm or refute a hypothesis.

  • But, would you think about controls?
  • Record the parameters you use?
  • Would you be able to replicate the results? The sequence database is constantly being updated, so it is likely you couldn’t replicate it.

When using bioinformatics methods reading the original paper will help you understand what it can be used for and, crucially, any limitations of the method.

Computational experiments still need controls, both positive and negative. Let’s consider a few of examples:

ExamplePositive controlNegative control
A method to predict protein structureA sequence of a protein you know the structure off from x-ray crystallographyA random protein sequence
A method to identify tissue type from histology imageAn image of known tissue typeAn image of a blank slide
A method to predict gene structure from DNA sequenceA sequence of a gene with known structureA scrambled sequence

Some methods will always give an answer, regardless of its accuracy – it’s useful for you to know that kind of detail about a method! So it is important to read about the method prior to using it.

After thinking about controls for using these kinds computational methods, are there any analyses you have done previously that you would do differently now?

The application of bioinformatics involves using other people’s methods and sometimes creating your own. On the next page, we’ll discuss whether or not learning programming is essential.