MMM-D: database of co-evolving proteins
Home FAQs Contact us

FAQ
Enter protein ID:
Select database:



  1. What is MMM-D?
  2. What is Match Potential?
  3. What is MMM score?
  4. What is MMMpG score?
  5. Which sources of known interactions were used in developing MMM-D?
  6. How many proteins are there in MMM-D?
  7. How many protein-protein interactions (PPIs) are there in MMM-D?
  8. What kind of protein ids have been used to name the downloadable files?




1. What is MMM-D?

MMM-D is a database of co-evolving proteins. It is called MMM-D because the co-evolution between proteins was detected by our previously developed tool MatrixMatchMaker (MMM). To read more about the method by which MMM measures protein coevolution via MMM score, please read The Human Protein Coevolution Network by Tillier, E. R. M and R. L Charlebois (2009).


All eukaryotic proteins in orthologous clusters in the OMA database (both Oct-2009 and Nov-2010 versions) were aligned with MAFFT v.6 then distance matrices were obtained using prodist from PHYLIP. MMM was run to find the largest common submatrix between pairs of matrices with at least 20 common species.
The tolerance perameter was set to 0.1 for these runs, corresponding to a threshold of distances not being more than 10% different between the two matrices.

Top


2. What is Match Potential?

The maximum possible MMM score. Low Match Potential values indicate that a high MMM score is impossible in principle for these inputs.
Since OMA clusters do not contain paralogs, the Match Potential in MMM-D is trivially the number of species in common between the phylogenetic trees of two proteins.

Top


3. What is MMM score?

The size of the largest common submatrix between the two distance matrices. The higher the MMM score, the stronger the coevolution, and consequently the likelihood of protein-protein interaction.

Top


4. What is MMMpG score?

The MMMpG score is the MMM score corrected for the overall phylogenetic signal that is present due to speciation rather than to interaction.

Top


5. Which sources of known interactions were used in developing MMM-D?

We consulted the following online available resources to separate the already known protein-protein interactions (PPIs) from MMM predictions (unknown PPIs). Information about the methods of interaction detection methods, source databases and taxon ids of the interacting proteins can be obtained by clicking on the Interactor # (1st column) in 'Known Interactions (green / white)' table.

BioGRID
ChaperonDB
CORUM
Costanzo et al., 2010
DIP
HPRD
Human Gold-standard dataset
IntAct
MINT
NCBI
PhosphoGRID
Worm Gold-standard dataset
Yeast Gold-standard dataset

All interactions from those databases were mapped to the OMA clusters in which the corresponding proteins were found, thus creating interologous interations between the orthologous proteins from other species.

Only those known interactions with MMM score of 3 or more are shown.

Top


6. How many proteins are there in MMM-D?

MMMD contains 13838 human and 1417 yeast proteins for OMA Oct-2009, and 20839 human proteins for OMA Nov-2010.

Top


7. How many PPIs are there in MMM-D?

Following is the number of PPIs of each kind in MMMD:

OMA Oct-2009:
Known (Human): 60277 (Strong:  182, Medium:  2192, Low:  15366)
Predictions (Human): 5325101 (Strong: 1727, Medium: 163445, Low: 5159929)
Known (Yeast): 32025 (Strong: 61, Medium: 898, Low: 6270 )
Predictions (Yeast): 71163 (Strong: 390, Medium: 8038, Low: 62735)

All pairs with MMMpG >=6 and MMMpG > MMM are shown.
All known interactions with MMM score of 3 or more are included.

Predictions are qualified as Strong, Medium or Low based on the MMMpG scores according to the frequency of known interactions for pairs above a certain MMMpG score.
For MMMpG > 14, known interactions are ~10% of the predictions (from the databases above and unpublished data)
For 9 < MMMpG < 15, known interactions are ~1% - 10% of the predictions (from the databases above and unpublished data)
For MMMpG < 10, known interactions are >~1% of the predictions (from the databases above and unpublished data)



OMA Nov-2010:
Known (Human): 323702
Predicted (Human): 358841

All pairs for which both MMMpG >=12 and MMM >= 12 (designated as MMM12+) are shown.
All known interactions with MMM score of 3 or more are included.

Top
8. What kind of protein ids have been used to name the downloadable files?

These are human and yeast protein ids from OMA database. Please be sure to use the appropriate OMA version, since the OMA protein IDs are not always preserved accross different versions.

Top