Skip to main content
Fig. 1 | BMC Bioinformatics

Fig. 1

From: doepipeline: a systematic approach to optimizing multi-level and multi-step data processing workflows

Fig. 1

Schematic visualization of doepipeline design space movement. Example of optimization of two factors (A and B) through both the screening (a) and the optimization phase (b), completed in 3 iterations. Each dot represents an executed pipeline with the parameters set by factors A and B. Triangles represent executed pipelines using the optima of an Ordinary Least Squares (OLS) model calculated in each optimization iteration. Red dots and triangles represent the best configuration of factors found in each iteration. Dashed lines represent the current high and low parameter settings in each iteration. Screening phase: a GSD using three levels and a reduction factor of 2 is used to span the design space. The pipelines are executed with the factor configurations suggested by the GSD and an approximate optimum is found (red dot). Optimization phase: in iteration 2, an optimization design is created around the best configuration found in the screening phase (black dots). In iteration 3, the design space is moved in the direction of the configuration of factors that produced the best result (red triangle) in iteration 2. doepipeline halts when the best response is produced by a configuration of factors that lies close to the center point (red triangle in iteration 3).

Back to article page