Notebook description

This Jupyter notebook analyses the performance of pyCapsid for a range of protein shells

Range of capsids analyzed

Memory usage versus protein shell size

The screening of the memory usage as a function of the protein shell size indicated a power law relationship (linear trend when plotting variables in a log-log plot).

Below we first calculate the linear regression model in the log-log space and then plot the data and the model.

Runtime versus number of residues

General check of runtime data

Exploratory plots

Assessment

Assessment:

Assessment:

Statistical analysis

Accuracy analysis

Each generated coarse-grained model simulation was compared with the experimental model. More specifically, the thermal fluctuations (captured by the B-factors) were compared using the correlation coefficient as a metric. Here we analyze how the correlation coefficient changed as a function of the protein shell size and the resolution of the experimental data. We focus on the results from the unified elastic network model (uENM), which is the default in pyCapsid.

Impact of capsid size

Evaluation:

Evaluation:

Impact of the empirical resolution of the capsid reconstruction

Evaluation:

Evaluation:

Benchmark with ProDy

Assessment:

Assessment: