Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: GenoGAM 2.0: scalable and efficient implementation of genome-wide generalized additive models for gigabase-scale genomes

Fig. 2

Coefficient estimation performance. a Empirical runtime for the estimation of coefficients vector β is plotted in log-scale against increasing number of parameters (also log-scale). The runtime is capped at around 2 hours, such that runtime of previous GenoGAM version (blue line) terminates after 1100 parameters. The new version of GenoGAM (orange line) achieves linear runtime in p (dotted line p), the number of parameters, compared to the previous cubic complexity (dotted line p3). b Memory consumption in MByte for the estimation of coefficients vector β is plotted against number of parameters (also log-scale). Due to the runtime cap at around 2 hours the runtime of previous GenoGAM version (blue line) does terminate after 1100 parameters. The storage of matrices in sparse format and direct solvers avoiding full inversion keep the memory footprint low and linear in p (dotted line p) for the new GenoGAM version (orange line) compared to quadratic in the previous version (blue line, dotted line p2)

Back to article page