Fios Genomics – Our Data Analysis Process
Articles Blog

Fios Genomics – Our Data Analysis Process


Data analysis at Fios is a three stage process. First thing we do is robust quality control.
We look at platform-specific quality, we look at general quality properties for instance
outlier detection. We use those as key metrics to isolate poor quality samples. For us, quality control is a pipelined process,
just to make sure that the data that’s going into the analysis is the best that it can
possibly be. It’s absolutely critical to spot problems
or artefacts, are the samples behaving the way they should? Is the data of a sufficient quality to actually answer the questions the client’s interested in? Stage two is the statistical analysis of the
data, comparing the groups of interest and pulling out the nuggets of information. We want to assign a p-value, a confidence, whether a gene for example is changing in a particular
scenario, or whether a protein is increasing in a certain category of people. The key thing is to use the right statistical
model to answer that question. Stage three in the process is perhaps the
most important. The interpretation of the data. What is it telling us about your drug, your
patient, your study. Most of the bioinformaticians working at Fios
come from a biology background. Often we’ll go to the pathway level or the gene ontology
level, and that really requires biological knowledge and I think that’s where Fios is
a little different. We know the questions the researcher is interested in, if that’s going
to be giving them the things they want to take actions on. [music]

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top