NetworkBLAST input and parameter setting

The accepted formats for PPI files are:

i) TAB – separated:

ID1 ID2 PROBABILITY_OF_INTERACTION
ID3 ID2 PROBABILITY_OF_INTERACTION2
.... 

where ID1, ID3 and ID2 are IDs of species proteins and PROBABILITY_OF_INTERACTION represents the reliability of the interaction (from 0 to 1). The interactions are undirected, i.e., ID1-ID2 is the same as ID2-ID1. The example below demonstrates 5 valid interactions in one protein network with varying reliability values:

# -------------BEGIN SAMPLE INTERACTION FILE------------
 
YAL001C YBR123C 0.4489
YAL002W YKR026C 0.4124
YAL002W YLR148W 0.5722
YAL002W YLR396C 0.5211
YAL002W YMR231W 0.5786
 
# --------------END SAMPLE INTERACTION FILE-------------

ii) PSI MI 2.5:

The new XML interaction data description scheme established by HUPO and adapted recently by BIND, DIP, HPRD, IntAct, MIN, BioGrid and other databases. The server was tested to handle successfully PSI-MI 2.5 recent files downloaded from DIP site.

In contrast to TAB-separated PPI files, in this case the interaction probabilities are calculated automatically (on the sever side) using the logistic regression method of Sharan et al., 2005. The computation of the reliability of an interaction is based on the types of experiments in which the interaction was observed.

Notes:

- Input file format is detected automatically. PPI files can be uploaded gzip-ed, decompression will be performed automatically.

- The server currently handles PSI files containing protein-protein interaction data only. Files containing interactions of other types
(see, e.g., BioGrid's), need to be preprocessed to remove these interactions prior to uploading.

BLAST data file should be in the format below:

ID1 ID2 EVALUE
ID3 ID4 EVALUE2
... 

where the first column should relate to species 0 and the second to species 1

For example:

# -----------------BEGIN SAMPLE BLAST VALUES------------
 
YCR065W CG10002 2e-18
YIL131C CG10002 1e-17
YNL068C CG10002 5e-18
YPR104C CG10002 7e-12
YIL023C CG10006 7e-19
YCR051W CG10011 2e-08
YDR264C CG10011 4e-12
YGR232W CG10011 3e-10
YGR233C CG10011 2e-09
YIL112W CG10011 5e-16
 
# ------------------END SAMPLE BLAST VALUES-------------

Note: Input file can be uploaded gzip compressed, decompression is performed automatically.

Parameter description:

• Density of a complex: Probability of an interaction within a complex. Higher values result in denser complexes.
• BLAST threshold: E-Value threshold to consider two proteins as sequence similar (and potential orthologs).
• False negatives: Expected rate of missing interactions in the corresponding PPI file.
• Reliability computation threshold: This parameter is relevant for PSI-MI input files only. It controls the features of the logistic regression employed by the server for assessing the reliability of each interaction. Specifically, it sets the threshold for considering a PPI detection experiment as large-scale and allocating it a separate feature vs. considering it small-scale and aggregating all small-scale experiments into one feature.

For more details on the parameters, please see Sharan et al., 2005