| Poster | Help | Intro | logoMEM - Multi Experiment Matrix
P. Adler, R. Kolde, M. Kull, A. Tkatšenko, H. Peterson, J. Reimand and J. Vilo: Mining for coexpression across hundreds of datasets using novel rank aggregation and visualisation methods (2009) Genome Biology [abstract]
R. Kolde, S. Laur, P. Adler and J. Vilo: Robust rank aggregation for gene list integration and meta-analysis (2011) Bioinformatics [abstract]
Enter gene ID(s) (for example: Jun, 203325_s_at, ENSG00000204531, ...) [?] Database: [?]
Select collection [?]  
to browse all datasets in our collection click here
Options:  Similarity   Output   Gene filters   Dataset filters   Reset session 


TBA - Major update:
  • We use cookie now to remember user session. You can leave the page, return later and continue from where you left off.
  • We changed the web layout a bit. Some links were rearranged to make it more logical and more easy to find (let us know if you miss something).
  • New visualisation of g:Profiler results (via g:P biojs plugin) is displayed right under the matrix to give instant overview of enriched functional GO terms for the found gene list.
  • Expression data database has been updated (and will again soonish).
  • We also made public first iteration of processed RNA seq data for more popular organisms. Due to large amount of genes not expressed anywhere there can be some unpredictable behaviour of rank aggregation. We'll keep working on that!
  • We have moved to larger server and thus the stability and capasity of the site are improved.
  • Gene versus gene view is reimplemented
22.04.2014 - Minor update to fix some rare crash bugs and correct HTML. Separated filter for unknown and ambiguous genes, now they can in-/excluded separately. Updated MEM cpp backend.
17.04.2013 - We have improved MEM cpp backend. Computationally intense steps are sped up by using multiple CPU cores.
26.09.2012 - It is now possible to download MEM results in NetCDF format. You can find download link under 'Query details'. For NetCDF structure and tutorial see Help. Added reference to MEM methods paper about Robust Rank Aggregation: Link to R package.
18.05.2012 - We have updated our database with standard RMA normalization version that is used by default. Also a version based on custom CDF mappings from BrainArray using reference FARMS normalization was added. All mappings are targeted on ENSG identifiers. You can choose the database version from respective menu.
10.01.2012 - Text search for dataset selection added under Dataset filters tab. Try it out!
02.11.2011 - Password protected database capability added to MEM! Contact us for more information!
02.02.2011 - We have updated our expression experiment database from ArrayExpress repository. Please note that default (i.e Current) database version refers now to the latest version (26.12.10).
01.12.2010 - Default dataset limit set to 100 - no more than 100 datasets are used in initial query for speed purposes, this parameter can be changed under "Dataset filters" tab. New parameters added under "Output" tab to manipulate cell size and spacing in visual output.
26.04.2010 - Arabidopsis Genome [ATH1-121501] and Rice Genome Array [Rice] platforms are supproted by g:Convert now. Newer and larger database version (20.12.09) has now set as default.
15.02.2010 - First update after publication; New feature - "Database version", which includes updated version of ArrayExpress gene expression dataset repository. [read more]


MEM is a web-based multi experiment gene expression query and visualization tool. It gathers several hundreds of publicly available gene expression data sets from ArrayExpress database. Different data sets feature different tissues, diseases and conditions. For better compatibility and comparability data sets are arranged by the platform type.

Given a gene as an input, MEM ranks other genes by their similarity in each individual data set. The essence is a novel rank aggregation method that takes those individual rankings and comes up with a score of significance and hence a ranking across all datasets simultaneously.

The new significance score is also capable of identifying a subset of data sets where the genes are significantly similar, thus allowing to eliminate those where the correlation is missing or not detectable.

MEM quick-start:

  1. Input a gene name into text field (i.e. Jun, 203325_s_at, ENSG00000204531, ...)

  2. (optional) From drop-down menu select microchip platform to be used in the query. Platforms are sorted based on number of experiments available in our database. All experiments are downloaded from ArrayExpress.

  3. Then hit "Submit query" button.

  4. If gene has more than one matching Affymetrix probeset id, then the selection will be prompted just above platform selection. Only one probeset id can be selected, then hit "Submit query" button again.


  • Example query 1
    Query of COL5A1(Collagen alpha-1(V) chain precursor) on Affymetrix GeneChip Human Genome HG-U133A platform. Note how clustering divides datasets into two large group. ( ~ 1 m, 18 s)
  • Example query 2
    Query of NANOG (Early embryo specific expression NK-type homeobox protein) on Affymetrix GeneChip Murine Genome 430 2.0 platform. Standard deviation of query is greater than 0.29 and only used datasets are shown. ( ~ 20 s)
  • Example query 3
    Query of a known breast cancer associated gene BRCA2 on Affymetrix GeneChip Murine Genome U74Av2 platform. Standard deviation of query is greater than 0.29 and all datasets are shown. ( ~ 10 s)
  • Example query 4
    Query of yeast gene UTP8 (U3 small nucleolar RNA-associated protein 8) on Affymetrix GeneChip Yeast Genome S98 platform. ( ~ 6 s)

Tips & Tricks:

  • [?] MEM output explained
  • Most of the standard gene/probeset/protein identifiers for different species may be used in uploaded datasets. IDs are always treated in case insensitive manner.
  • Links like [?] provide pop-ups with MEM help and documentation, clicking will direct to full help page.

Contact :

Name *
E-mail address *

Please type contents of the above image *
Message *

Fields with * are required for form submission.

Priit Adler & Raivo Kolde & Jaak Vilo

Bioinformatics, Algorithmics, and Data Mining group BIIT
Institute of Computer Science
University of Tartu
Liivi 2-314, Tartu 50409, Estonia

Estonian Biocentre
Riia 23, Tartu 51010, Estonia


Multi-Experiment-Matrix © 2008-2016 | Priit Adler & Jaak Vilo @ Biit Group, Institute of Computer Science, University of Tartu
Sat Feb 25 18:53:04 2017
Sat Feb 25 18:53:05 2017
Duration : 0 m, 1 s