Domain knowledge is important in rare disease data analysis:
In addition to methodological considerations, collaboration with domain experts may result in unexpected insight into potential sources of variation. As an example, consider a study of neurofibromatosis type 1 (NF1) datasets.[@doi:10.3390/genes11020226] These datasets were, unbeknownst to the computational biologists, generated from samples obtained with vastly different surgical techniques (laser ablation and excision vs standard excision), resulting in substantial biological differences that are a consequence of process, not reality. One might expect, in this example, that this technical decision would result in profound changes in the underlying biology, such as the activation of heat shock protein related pathways, unfolded protein responses, and so on. Consequently, careful assessment of and accounting for confounding factors is critical to identifying meaningful features within a dataset.
Domain knowledge is important in rare disease data analysis:
Can strategies/methods exploiting "anchoring" help rare disease data analysis [https://www.cell.com/cell/fulltext/S0092-8674(19)30559-8] ? Future direction?