BerensF / ComparingProteins

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ComparingProteins

Compares proteins by there isosufaces and clusters them.

Call:

./CompareIsosurfaces.R _parameterfile_

Parameterfile

The parameterfile contains following parameters:

Name Parameter Description
PathToData .../ProteinComp/Input/ Path to the protein data, every protein in it own folder that contain the .dx file
PathToOutput .../ProteinComp/Output/ Path to the Outputfolder
n 100 How many points will be selected in every round
m 500 How often the calculation will be done
PathToCPProgram .../ComparingProteins/LowerBounds/FirstLowerBound/main Path of to the Cpp program on your machine
PathToRProgram .../ComparingProteins/EMDandClustering/AllLowerB_EMD_Clust.R Path of to the R program on your machine

Comments can be added by a '#' in the parameterfile. The parametername and parameter have to be seperated by a '='

Folderstructure

└─── ProteinComp
│   └─── Input
│       │   └───  Prot01
│       │         │   Prot01.dx
│       │         │   ...
│       │   └───  Prot02
│       │         │   Prot02.dx
│       │         │   ...
│       │        ....
│       │   └───  Protk
│   └─── Output
│       │   └───  Prot01_negative_n
│       │         │   Prot01_Prot01_negative_n
│       │         │   Prot01_Prot02_negative_n
│       │         │   ...
│       │         │   Prot01_Protk_negative_n
│       │   └───  Prot01_positive_n
│       │         │   Prot01_Prot01_positive_n
│       │         │   Prot01_Prot02_positive_n
│       │         │   ...
│       │         │   Prot01_Protk_positive_n
│       │   └───  Prot02_negative_n
│       │         │   Prot02_Prot02_negative_n
│       │         │   Prot02_Prot03_negative_n
│       │         │   ...
│       │         │   Prot02_Protk_negative_n
│       │        ....
│       │   └───  Protk_positive_n
│       │         │   Protk_Protk_positive_n
|       |  ListofEMD_negative_n
|       |  ListofEMD_positive_n
|       |  Dendrogram_UPGMA_Max
|       |  Dendrogram_UPGMA_Mean
|       |  Dendrogram_UPGMA_Neg
|       |  Dendrogram_UPGMA_Pos

Lower Bounds

There are three lower bounds of the Gromov-Wasserstein distance in this repositories. The first lower bound has a time complexity of O(n^2) the second O(n^4) and the third O(n^5). It is recommended to use the first lower bound, since it is much fuster than the other two. For more details see "Quantitative comparison of protein isosurfaces with approximated Gromov-Wasserstein-distance" from Felix Berens or "Gromov-Wasserstein Distances and the Metric Approach to Object Matching" from Facundo Mémoli.
For the third lower bound a linear program has to be solved. This is done with the library lpsolve_5.5, which can be downloaded here https://sourceforge.net/projects/lpsolve/files/lpsolve/5.5.2.0/lp_solve_5.5.2.0_source.tar.gz/download

FAQ

What can I do if i forgot a protein structure or have a unneeded structure in the Input folder?

Just add the forgoten protein structure or delete the unneeded structure and rerun the program. The program autmatical recognises, which structures were already compared.

I am only interested in the lower bound values of two specific protein structures (for example: Prot01 and Prot02), what can I do?

For this you already need the files containing the coordinates on the isosurfaces (here those four: Prot01_pot_neg.pts , Prot02_pot_neg.pts , Prot01_pot_pos.pts , Prot_02_pot_pos.pts). Then you can start the lower bounds with the following call from the terminal: ./main inputfile1 inputfile2 n m So for example if you are in the folder ProeinComp and want to compare Prot01 and Prot02 with 100 points each 500 times, you have to call those two lines:

./[pathToComparingProteins]/ComparingProteins/LowerBounds/FirstLowerBound/main Input/Prot01/Prot01_pot_neg.pts Input/Prot02/Prot02_pot_neg.pts
./[pathToComparingProteins]/ComparingProteins/LowerBounds/FirstLowerBound/main Input/Prot01/Prot01_pot_pos.pts Input/Prot02/Prot02_pot_pos.pts

About


Languages

Language:C++ 70.8%Language:R 25.9%Language:Batchfile 2.3%Language:Makefile 1.0%