Performance evaluation

The performance of Tentacle has been evaluated on its scaling characteristics when running in a distributed enviroment, as well as the quantification accuracy of the abundance estimation features. All evaluations and tests are described in the article (see Citing Tentacle).

Scaling performance

Tentacle scales very well with increasing computing resources. The following figure shows how the throughput of Tentacle scales when increasing the number of utilized worker nodes. For complete test details, refer to Citing Tentacle. The evaluation results data and code to generate the figures are available for download from figshare. A non-interactive version of the IPython notebook can be viewed in your browser.

Tentacle scaling

Tentacle scales very well with increasing number of computer nodes.

Quantification accuray

The accuracy of quantification and coverage was also estimated. The following two figures show the expected and measured coverage and quantification results. For complete test details, please refer to Citing Tentacle and the IPython notebook which contains the code to generate the figures. The evaluation results data and code to generate the figures are available for download from figshare. A non-interactive version of the IPython notebook can be viewed in your browser.

Coverage and quantification accuracy

Tentacle achieves very good coverage and quantification accuracy.

Distribution of measured quantification results on a semi-synthetic test case.

The distribution of the measured quantification levels follows the expected values for this synthetic test case.