Back to "Biological Language Modeling Seminar Topics"

Back to "Introduction to Protein Protein Interactions"

Overlap in Protein-Protein Interactions Datasets

Yeast-2-hybrid datasets:

from Deane et al 2002. All data are yeast-2-hybrid datasets.

_EPR, corresponds to the fraction of the true positives in the experimental dataset. Example: The _EPR is calculated as 31 ± 3% (Table II) for the GY2H data, suggesting that 70% of the reported pairs in this set are, in fact, false positives.

TABLE II EPR index

EPR index, _EPR, calculated for several subsets of DIP-YEAST (see "Results" and "Experimental Procedures" for details) using INT and RND1 subsets as representative for the interacting and noninteracting protein populations, respectively, is shown. The values of ² and N, the number of degrees of freedom, are given.

 

Dataset EPR 2 N DIP-YEAST 0.48 ± 0.03 9.07 29 EC2 0.85 ± 0.06 1.65 16 EC3 0.88 ± 0.17 3.05 10 GY2H 0.31 ± 0.04 14.84 29 GY2H' 0.50 ± 0.03 14.09 29 PVM 0.78 ± 0.13 5.85 16 CORE 0.92 ± 0.03 1.69 19 ITO1 0.22 ± 0.06 19.4 29 ITO2 0.41 ± 0.11 12.6 19 ITO3 0.58 ± 0.11 10.1 16 ITO4 0.62 ± 0.16 9.5 14 ITO5 0.55 ± 0.18 8.8 14 ITO6 0.57 ± 0.24 7.1 12 ITO7 0.57 ± 0.32 6.0 10 ITO8 0.65 ± 0.42 4.6 7

Mass Spec Datasets:

Gavin et al.:

7% overlap with yeast-2-hybrid datasets

56% overlap with YPD database (known complexes)

[yeast-2-hybrid has 10% overlap]

25% of all proteins are covered in Gavin study, while 95% of all proteins are covered in Yeast-2-hybrid system

Ho et al:

Table 1. in HoNature2002.pdf