Please read README-MSB.txt first. Code provided here is the latest version. Following additional data are found here:

* APC contains apcluster.m

* Clustering contains all code used for clustering similar patterns:
libsvm_gridsearch.m contains the code for optimizing the SVM training parameters
livsvm_normalize.m contains the normalization for SVM training
run_apc.m converts the TIs and runs apcluster.m or the C-code (in Utilities in the Supplemental Data)
sc_apc2svm.m is the script that generated the SVM training set or each cluster

* GO contains code for the GO-analysis:
amigo_enriched.pl was used to fetch the enriched GO terms from the Amigo web-site
go_dotplot.m contains code to generate all of Figure 7
my_GO.m contains the SQL queries for the GO-database (used with pg_query.m)
sc_parseamigo.m is a script to parse the output of amigo_enriched.pl into Matlab

* MYSQL_insitu contains a dump of an excerpt of the current insitu database. Only tables relevant for the TI queries were retained. Results may not be completely identical to the results in the paper because the dump is from a more current version of the data. Future updates will be on the main web-site www.fruitfly.org

* Visualization contains additional visualization tools for the TIs:
disp_moverlay2.m is an investigative tool to compare a set of TIs to a reference pattern. Displayed are all panels with the TIs, the reference patterns and the overlays of both. 
disp_mcomposites2.m is a more powerful workbench and investigative tool to compare a set of TIs to a reference pattern. It displays the consensus, allows for selection of individual TIs and for the selection of triangles. If a triangle is selected, only TIs with expression in that triangles will be displayed. 

All code is provided AS IS. For questions, please Email Erwin Frise erwin@fruitfly.org