The E Coli Predicted Protein Interactions Database.[Home] [Info] [Contact] [Search Help] |
|
This system is described in:
The "in silico 2 hybrid system" to predict interaction partners from sequence information is based on the search for pairs of positions between two proteins that show a correlated mutational behaviour that could be due to space neighbouring between those two positions. The positions mutational behaviour is extracted from multiple sequence alignments.
The database contains predictions for 67238 pairs of proteins (1% of all posible
pairs) which comprises predictions for 430 proteins (10% of the E coli
proteins). This is due to the neednees of a minimum number of corresponding orthologous
sequences to make the multiple sequence alignments. Homologous sequences
to E. coli proteins were taken from 14 fully sequenced bacteria. In order for
a pair of proteins to be evaluated, the two proteins must be present in the same
subset of, at least, 11 genomes.
Each calculated pairs of proteins has an asociated score.
Please note that the data contained here are . Do not take these data as experimental results.
You can access the multiple sequence alignments used for the calculations here (PIR format). These alignmens were generated from 14 fully sequenced bacteria in November 1998.
We have run the system again starting with a set of 48 genomes and ended up with calculations for more than 1/2 million pairs. We also implemented other prediction methods. The results will be available soon.
See [Search Help] for information of how to use the database.
NOTE: Each machine can make a limited number of queries in order to avoid abusive usage. If you reach that limit and you receive the corresponding error message, please contact us and we will restore your access.