Lawrence Berkeley National Laboratory

The sequencing process is a multi-stage operation requiring a combination of resources, machines, and operators and resulting in a series of DNA sequences. In a DNA sequencing production line, there exist a network of paths in which any single traversal yields a sample. Within this multiplicity of paths, specific combinations of steps contribute to under-performing or otherwise anomalous results. Isolating these combinations is difficult as many are not actualized and direct analysis quickly becomes computationally intractable. To solve this problem, both parametric and non-parametric techniques are used to autonomically isolate anomalous traces. Then, through a novel variation on principal components analysis, the contributing modi operandi are identified.