About ConStruct

ConStruct is an RNA alignment editor and consensus structure prediction tool. It combines multiple sequence alignment, thermodynamic structure prediction and statistics in a semiautomatical fashion. Its sophisticated GUI guides the user through correcting an initial sequence alignment with respect to a consensus structure. Its built-in structure prediction routines allow for:


  1. Create an intial alignment using your favourite tool (e.g. Clustal , Proalign , MAFFT etc.)
  2. cs_fold (ConStruct RNAfold front-end) creates thermodynamic base pairing matrices for each sequence
  3. Gaps from the alignment are inserted into the matrices
  4. The Mutual Information Content/Score (MI) is computed as measure for compensatory base pair mutations
  5. ConStruct's GUI (cs_dp): The base pair matrices (step 2) are superimposed and displayed in a consensus dotplot (upper right triangle). The MI is displayed in the bottom left part. Futhermore an alignment editor is displayed. Upon changes in the alignment, the consensus dotplot is directly updated.
  6. Consensus structure prediction is done by a weighted (and filtered) linear combination of the thermodynamic consensus dotplot and the MI. Predicted structure can be viewed and stored in a variety of formats (Circles-plot, color-coded alignment, interactive squiqqles-like plot, connect, RNAML etc.)


The current version is 3.2.5.
You can either install ConStruct from source or by using a precompiled Debian package.

Installation from Source

Generic INSTALL instruction are included in the package tarball.
You will have to make sure that the following packages are already installed on your system: The easiest way to install those packages is to use the devel packages from your Unix/Linux distribution. Mac OS X users will need to install fink.

Next download the construct tarball
Unpack the tarball
$ tar xvfz construct-X.Y.Z.tar.gz
cd into the newly created directory:
$ cd construct-X.Y.Z
Configure, compile and install (as root) the package
$ ./configure
$ ./make
# ./make install
In some cases you will need to point configure to your and For example, on a Debian box you would have to use:
$ ./configure --with-tcl=/usr/lib/tcl8.4/ --with-tk=/usr/lib/tk8.4/

Installing the Debian Package

Download the Debian package (3.2.4) (i386 only) and type
# dpkg -i construct*.deb
If you don't have RNAfold installed, you may also want to use this Debian package (i386 only).

Installation problems

The following error is known to occur on some SuSE und Fedora machines when cs_dp is started:
"Error in startup script: can't find package drawstructcore"
In most cases this is caused by a wrongly namend library. The following procedure fixes the problem:


  1. Create a sequence alignment of your RNA sequences.
  2. Load this alignment in cs_fold
  3. Click
    • "Write Project File" (writes an ASCII-file for cs_dp),
    • "Fold" (calls RNAfold for each sequence)
    • and "Exit"
  4. Load the created project file (<alignmentfilename>.proj) in cs_dp
  5. You can move nucleotides by selecting them with your mouse and (left or right) clicking on the arrow button. Try to move helices (green diagonal) on top of each other. This way they are building a consensus helix (red diagonal).
  6. Adjust the calculation options (menu: "Calculation_Base") to your needs. For tertiary structure predicion you will have to set a reasonable high multiplier for the MI.
  7. Consensus structure prediction routines are called through the "Structure_Prediction" menu.
Please have a look at this short demonstration.
For some more details see the manual (manual.pdf)).


Supplementary data for Wilm, A., Linnenbrink, K. & Steger, G. (2008)


If you find bugs, need help, want to discuss results, don't hesitate to contact
Gerhard Steger

Other consensus structure prediction tools

The following tools predict consensus structures from (fixed!) alignments: