Cargo Cult Bioinformatics

The metaphor of "Cargo Cult" characterizes a process that goes through procedures that appear superficially correct, even impressive, but that cannot deliver anything of significant value; this may be because

the procedure is incomplete, thus the ultimate objective cannot be reached -

• an analysis of genes coregulated in inflammation might be made, claiming that this may help us to understand autoimmune disease: in reality there is no concept of exactly how that kind of knowledge might be integrated to inform us about systemic pathologies,

the procedure is defective in execution, thus the objective cannot be realized because the procedure never terminates -

• a measurement might be undertaken with an experimental setup that is not sensitive enough to detect the expected effect;
• missing positive and negative controls fall into that category, because they preclude the interpretation of the experiment's results

the procedure rests on a fallacious assumption, thus the objective has no relationship to the procedure, even if it is successfully completed -

• a measurement is undertaken of a variable that does not actually contribute to the effect one claims to be studying;
• a bioinformatics algorithm might be tested only on synthetic data, or might make assumptions that may be valid for a "toy problem" but that break down in the real world;
• a recombinant vaccine might be developed in a model organism where the failure to elicit a protective effect does not prove it would not work in humans, and success does not prove it either.
• an inferior algortihm for multiple sequence alignment might be used for homology modeling;

or some other *conceptual* deficiency.

Thus I characterize as Cargo Cult Bioinformatics processes that don't, won't or can't inform the starting hypothesis.