Skip to main content Skip to navigation

Secondary Structure Neural Network

SSNN, or secondary structure neural network is a self organising feature map (SOFM) used to cluster data from circular dichroism to estimate protein secondary structures developed by Vincent Hall (now in Computer Science), Anthony Nash (now at Oxford) and Alison Rodger (now at Macquarie).

A free download of the software is available for academic use only, but maybe be licenced for commercial use. For full terms and conditions see the licence (coming soon).

In the left panel: The SSNNGUI links are for SSNN pretrained, easy to use, with 53 reference set proteins (from CDPro and Dichroweb's reference set 7 with 5 extra spectra, and a second reference set without the extra 5).

The SSNN1_2 Mac has versions for OS 10.8.4 ("Mountain Lion"), 10.6.8 ("Snow Leopard"), Windows 7, and Windows XP. These downloads are fully trainable with training (reference) sets provided.

SSNN spectra and structures maps

These are the links underneath the left panel. There are two reference sets available to train SSNN with, they are included in all packages from Mac 10.8.4 to Windows XP. There is a 48 protein reference set ("CDDATA.48"), this is called reference set 7 on Dichroweb ( CDDATA.48 was supplied by CDPro is also a set with 53 proteins ("CDDATA.48+5"), the same 48 plus 5 more: 2 spectra extrapolated to 100 % alpha-helix regular, and 3 with 'other' structures. 'Other' is the structure category for anything that is not alpha-helix, beta-sheet or turns.

SSNN protein CD spectrum

These application downloads come with the relevant MATLAB Compiler Runtime (MCR) release 2013a, 2011a, and 2011b .zips or links to download and install. The MCR is needed for any machines without MATLAB installed, so you do NOT need a computer with MATLAB, just the MCR.

For SSNNGUI please use 'SSNNGUI Instructions'see left panel. For SSNN1_2 instructions just look inside the packages. The SSNN1_2 is for code developers, and needs the weights and structures folder. This folder name must be changed to today's date in the form of "Weights_and_Structures_28000_14-Mar-2014". I hope I've included the Weights_and_Structures in all package downloads, if not, see the Weights_and_Structures downloads, the one with the date is the same thing, but for SSNN1_2 for code developers. SSNN1_2 does not need a weights and structures file, it can make that itself, but one is included for running the program quickly without training.

This SSN1_2 version can be used to cluster any dataset, but currently the structures (output) size must be set to 6, so:

if there are not enough outputs, just add some zeros to the end of the data matrix in to make it up.


If you are interested in using the Software commercially, please contact Warwick Ventures Limited("WVL"), the technology transfer company of the University, to negotiate a licence.

Shell for Mac:

The shell to run SSNNGUI or SSNN1_2 in Terminal is in the download list, you should just open it an TextEdit or some other text editor, and change it to refer to the correct name of SSNN. Open the Contents by "right-clicking" and choosing show package contents, then go to Contents/MacOS/ and see what the name of the app in there is, copy this to the bottom of the shell where it says:

eval "\"${exe_dir}/\"" $args