Next: 3. Methods 1 Up: Prediction of RNA Previous: 1. Introduction

2. Materials -- Computer Equipment and Programs

The mfold package by Zuker and Jaeger [14,15,16] computes optimal and suboptimal foldings of an RNA molecule. There are two principal improvements over the older RNAFOLD program [21]. The first is the incorporation of up-to-date energy rules, and the second is the program's ability to compute both optimal and suboptimal foldings. There are two forms of output. One is a list of optimal and suboptimal foldings, sorted by energy. The other is a graphics output that depicts all suboptimal foldings in a single image. The graphics output is called the energy dot plot.

Mfold was first developed in a VAX/VMS environment. This version is now obsolete, although it is still available. It will be referred to as the VAX version, the VMS part being understood.

The official version of mfold now runs on UNIX workstations and larger computers. My personal version runs on an Indigo₂ workstation (IRIX 5.2) by Silicon Graphics and will be called simply ``mfold''. It uses the SGI graphics library and is hardware specific. An X11 version of mfold also exists on a number of platforms. It is called ``xmfold'' on the IRIS, ``mfold-dec'' on DEC Ultrix workstations, and ``mfold-sun'' on SUN workstations.Ther are also incomplete versions that run on CRAY and CONVEX supercomputers, as well as an Apple Macintosh version (MULFOLD). A GCG [22] compatible version is now available.

At the heart of mfold are the two programs ``lrna'' and ``crna''. They fold linear RNA and circular RNA molecules, respectively. Computer storage requirements increase as the square of the sequence length. Folding times increase as the cube of the sequence length. On an Indigo₂ workstation, it takes 3.9 minutes to fold 500 bases. Space requirements (both CPU for run time, and disk for save sets) are roughly 4.1n² bytes for lrna and crna. It is usually space limitations that set the upper bound for the maximum size of an RNA that can be folded.

The interactive energy dot plot that is described below requires a graphics interface. In the VAX version, this requires a Tektronix 4105 terminal or terminal emulation, or a Visual 600 series terminal. The VAX version also requires a Tektronix PLOT10 object library. The VAX graphics implementation is quite primitive. The Silicon Graphics mfold version will run on any SGI workstation, but the user must be working on the color monitor to use the dot plot feature. The X11 versions all require an X11 library to be installed on the workstation, although the program can be run from any X terminal. The SUN version also requires the gnu C compiler. A hard copy of the energy dot plot can be produced with the package even if the interactive energy dot plot is not available.

The CRAY, CONVEX and Macintosh versions of the program have been ported without the dot plot. All versions of the program can be run from a non graphics terminal or in batch mode if the dot plot option is not selected.

The programs are written in FORTRAN using DEC extensions. Some of the UNIX versions have been altered to remove unsupported FORTRAN extensions. Two ``include'' files, lin.inc and circ.inc are provided with the source code for lrna and crna respectively. The value of the ``nmax'' parameter is the maximum fragment size that can be folded. It can be changed and the programs recompiled using the Makefile that is provided.

The VAX and UNIX workstation versions are available via anonymous ftp at snark.wustl.edu. Compressed tar files are in the pub directory and are named according to the descriptions above.

  Next: 3. Methods 1 Up: Prediction of RNA Previous: 1. Introduction

Michael Zuker
Thu Nov 2 14:28:14 CST 1995