First steps
Below you find a 5-minutes introduction to the graphical user
interface of ProCope. Follow the steps to learn how to load a set of
purification data, generate a scores network, cluster this network to
get a set of predicted complexes and finally evaluate the quality of
these complexes.
Starting the program
Start one of the start-up scripts in the main directory of ProCope, depending on your system:
gui.bat
for Windows users
gui.sh
for Linux and Mac OS users
Loading a set of purification data
- Right-click the purification data list at the bottom of the window, choose Load purification data from file. In the file dialog select a purification data file (which may be GZIPed). You will find purification data sets in the
data/purifications/
directory of the ProCope package.
- Alternatively, you can drag&drop a purification data file from your file browser directly into the purification data list.
Generating a scores network
- Select the purification data set you just loaded, then right-click it and choose Calculate scores.
- Choose Socio affinity scores and leave the cutoff field empty, click OK
- The scores network you just generated will appear in the networks list at the top of the window.
Clustering the network to predict a complex set
- Select the scores network you just generated, right-click it and choose Cluster
- Select Hierarchical agglomerative clustering, click OK
- Choose UPGMA and a cut-off of 1.0
- The clustering will appear in the complex set list in the middle
of the window. Note: We did not do any parameter tuning or
preprocessing steps. The predicted complexes are most probably not of
high quality.
Evaluating the complex set using a reference complex set
- Right-click the complex set list, choose Load set from file and load the MIPS reference set (
data/complexes/mips_complexes.txt
). Alternatively, drag&drop this file directly into the list.
- Select both the reference set and the clustering, right-click them in the list and choose Comparison => Brohee comparison. This will calculate the sensitivity, positive predictive value and accuracy according to Brohee et al., 2006. You can investigate the results.
- Next, choose Comparison => Map complexes. Select Only one mapping per complex. click OK. This time you will see which complexes of your prediction are mappable to which complexes of the reference.
Evaluating the complex set using localization data
- Select the menu Tools => Localization data. In the window which just opened, click Add localization data.
- Choose the localization data set by Huh et al., 2003 (
data/localization/huh_loc_070804.txt
).
- Close the window, select your clustering, right-click it and choose Quality => Colocalization
- In the dialog, select the Huh data set and the Colocalization score, click OK
- In the next dialog, simply click OK
- You will now get a list of the complexes of your prediction,
along with their colocalization scores and the overall colocalization
score of the clustering.
Evaluating the complex set using GO annotations
- As the GO files contain primary SGDIDs whereas the purification
data sets (and thus your predicted complex set) contain systematic
names, you need to use a name mapping set.
- Select the menu Tools => Name mappings, click Add name mapping.
- Choose the yeast name mapping file coming with ProCope (
data/yeastmappings_YYMMDD.txt
). Leave the target identifier setting as it is. Click OK and close the window.
- Now we add a GO setting. Select Tools => GO settings. Click Add GO setting.
- Select the
.obo
file from data/go/
as the GO network file and the .sgd
file as the annotation file. Leave all other settings as they are. Click OK. Close the window.
- Select your clustering, right-click it and choose Quality => Semantic similarity (GO).
- Select the GO setting you just created. in the next dialog, simply click OK.
- You get a list of your predicted complexes and their GO semantic similarity scores.