The codebase of the paper: Between-subject prediction reveals a shared representational geometry in the rodent hippocampus. The main idea of this paper (which was started at 2018 MIND hackthon) is to reveal a shared relationship in ensemble hippocampal (place) cells activity between different experimental conditions (left and right running trials on a T-maze in this case) across subjects using hyperalignment.
Clone and check out the hyperalignment branch of the vandermeerlab main codebase.
Before you try to do anything below, edit set_hyper_path for path setup (corresponds to your local path), and getTmazeDataPath and getAdrDataPath in utils/data folder for data setup.
Then execute set_hyper_path in your matlab command window.
Each figure has a corresponding script in
fig_scripts folder and data preparation script used in fig 2, supp. fig 2, fig 3 and supp. fig 3 is located at scripts/prepare_inputs.
Note that they might share the common scripts with only different parameter used (for flexibly testing different data input). For example, Q in figure 4 includes interneurons, so cfg_data.removeInterneurons = 0;, but figure 5 and so on excludes interneurons (cfg_data.removeInterneurons = 1;).
You should adjust the parameters before preparing the data input. (We might finalize the data input used then this is no longer required.)
The following tools/code were used in this paper. We thank all of their contributors.
shadedErrorBar: https://github.com/raacampbell/shadedErrorBar
hypertools: https://github.com/ContextLab/hypertools-matlab