MIST
MISTlogoMIST
v4.0 (June 2019)

How to use MIST
MIST Flow Diagram
Input Examples
Download Result Options
Cases at MIST
About MIST
Interactions at MIST
API
Contact Us

How to use MIST

MIST Flow Diagram

Screen 1: Input/Home Page

Home Page

Screen 2: Result Page

Result Page

Input Examples

Input examples for each species are available here.

Download Result Options

To download a network, right click on the network. An option to download a png/jpg or cytoscape js file will pop up. To download individual tables, click the “Export table” button on the top of each table. Alternatively, you can click the “Export all tables” button below the network to export all tables at once as .txt format.

Cases

Overview of possible outcomes when you enter a gene or protein list at MIST

CASE 1: (Interactions found in MIST):

You input a given a gene symbol or ID (or a list of them), or pairs, and all genes or pairs have interactors in MIST. In this case, MIST will find the interactions and display them.

Example: For Human gene input- 81570,10994

CASE 2: (No interactions found):

For all or a few of your input gene symbols or IDs, no interactions are present in MIST, so no interactions are shown. The inputs will show up as green nodes without edges. These inputs will also be listed in the table “no interactions found”.

Example: For Drosophila gene input in ppi dataset- CG43295

CASE 3: (Gene symbol or ID is not found, e.g. mis-spelled):

If your input gene symbol or ID is misspelled or otherwise not in the database, then MIST will not find any matches. In this case, no interactions will be found, and you will see a red bar on the top saying following genes not found. If there is a mix of ‘good’ IDs and IDs that are not found, then results for the good IDs will be displayed, and the ‘bad’ IDs will be noted at the red bar on the top.

Example: In any case input - foo

CASE 4: (Old ID with new mapping):

If your input gene symbol or ID has been split into two genes in a more recent annotation of the genome (e.g. Gene A is now called Genes B and C), then both of the IDs that map to your input ID will be displayed (e.g. MIST displays results for B and C). However, if one of the newly annotated genes has the original ID, then MIST will only use this ID (e.g. Gene A is now called A and D, and MIST displays results for A).

Example: For Human gene input - A1B

About

The Molecular Interaction Search Tool (MIST) is a comprehensive resource of molecular interactions. MIST currently supports several species, including fly, mouse and human. At MIST, you can mine known physical interactions and infer interactions using other supportive evidence as well as similar genes by correlation analysis. The web interface allows users to retrieve interacting or similar genes in table format as well as visualize these interactions as networks.

MIST currently supports search of information for 10 species:

Types of interactions displayed at MIST

Protein-protein interaction (PPI):

“PPI” refers to physical interactions between two or more proteins.Both binary and complex-based PPI data are available. This information was compiled by integrating experimentally identified PPIs as annotated in BioGRID, IntAct, mentha, DIP, DroID, HPRD, FlyBase, PomBase.

Interologs (PPI):

“Interologs-ppi” are interactions predicted based on PPI data obtained for orthologous proteins. For example, a pair of Drosophila proteins would be considered interologs-ppi (i.e. they are predicted to interact) if the human orthologs of these proteins were found experimentally to interact with each other. We annotated interologs as follows. First, experimentally identified PPIs were compiled from major PPI databases. Next, the relevant proteins were mapped to specific species using ortholog annotation from DIOPT database. In case of one to many ortholog mapping, we selected the best ortholog gene using the DIOPT score as the indicator of highest-confidence ortholog relationships (Hu et al., 2011). In addition, we also filtered out any orthologous with low rank and/or score below 3.

Genetic Interactions:

Genetic interactions indicate that the effects of mutations in one gene can be modified by mutation of another gene. Genetic interactions were collected from BioGrid, IntAct, DroID and FlyBase. Interologs (genetic interaction): “Interologs-GI” are interactions predicted based on genetic interaction data obtained for orthologous proteins. Orthologous mapping for interolog-GI is the same as interolog-ppi (see above).

Kinase-substrate interactions:

Kinase-substrate interactions are defined here as inferred interactions between a kinase and its potential substrate(s). We selected experimentally verified phosphorylation sites. Then we used the NetPhorest program to predict what kinase phosphorylated the site. This program is based on probabilistic sequence models of linear motifs to predict kinase-substrate relationship.

Phenotype correlation:

A phenotype correlation network was built based on gene pairs that share the similar or opposite phenotypes in cell-based functional genomic screens (Vinayagam et al., 2014). Very often, gene pairs that show significant phenotype correlations are part of same protein complex or involved in similar biological processes. A positive correlation suggests an activation-type relationship between the genes; a negative correlation suggests an inhibitory relationship between the genes.

Phospho-correlation:

This network was built on a data set created as follows (Sopko et al. 2014). First, specific kinases were disrupted in vivo. Next, changed phosphosites present in perturbed backgrounds vs control were identified. Then, a phospho-correlation network was built based on protein pairs that share similar or opposite pattern of phosphosites in kinase-deficient phosphorylation profiles. We have previously shown that such phospho correlation exist between functionally related proteins, e.g. they can be used to infer a kinase-substrate interaction (Sopko et al 2014). MIST contains phospho correlation relationships for Drosophila and S.cerevisiae.

Gene expression correlation:

We define expression correlation as the correlation of mRNA expression levels across many experimental conditions, cells and tissue types. MIST uses expression correlation information from DGET. Gene pairs with Pearson correlation coefficient >= 0.85 with at least 3 datasets are displayed.

Additional information about data sources

SourceVersion or date of dowloadInteractions included in MIST (vs3)Interactions included in MIST (vs4)IncreaseNon-overlapping interactions (vs3)Non-overlapping interactions (vs4)Interaction typesSpecies for MIST
DIPdip20170205106,660106,66006,517921PPI10
DroID (including DPiM)v2018_8247,816280,61713129,219128,632PPI,GI1
BioGridvs3.5.1711,827,2312,114,921161,105,7221,198,044PPI,GI9
IntAct (including MINT)5/23/2019634,547764,3342016,38156,505PPI,GI10
FlyBaseFB2019_264,00778,6612321,1922,881PPI,GI1
HPRDRelease976,23376,233023,36322,743PPI2
PomBase6/22/20175,2905,29003,6243,495PPI1
mentha5/13/20191,020,3511,202,463189,45412,108PPI9
HumanMAPKmanuscript table (2010)4,5304,53002,9412,911PPI1
WormBaseWS270NA53,078NANA29,999PPI,GI1
MIST (without interologs)vs42,376,3412,737,46615NANAPPI,GI10
MIST (including interologs)vs413,573,89716,441,35621NANAPPI,GI10

API

The Mist data can be queried programatically. To use the api use this end point

Contact Us

DRSC/TRiP Functional Genomics Resources

Department of Genetics

Harvard Medical School

New Research Building, Room 336

77 Avenue Louis Pasteur Boston, MA 02115

Email: perrimon@receptor.med.harvard.edu