History of Image

  • the first ever version of a program like Image was written by John Sulston et al. in FORTRAN to run on a VAX machine. (CABIOS Vol.4. no 1. 1988 Pages 125 - 132)

    The programs were written in Fortran to run on a VAX computer under VMS. At this time only a few programs had graphical interfaces and it was hard to correct mistakes. Nevertheless, large parts of the C.elegans genome were mapped using this software.

    Even the latest version of Image3 will still carry some historical legacy from this software from some 15 years ago.

  • In 1993 the first approach was made to move the image processing on to the UNIX machines. Richard Durbin wrote a little program image.c which displayed a rotated .dat-image, and superimposed the bands which were read from a .fea-file (feature-file). Those files were still created by the fortran programs on th VAX and then imported onto the UNIX system for testing. This early attempt also had a somewhat crude TAR-mode for editing the sample bands.

    The rather generic name given to that file ("image.c") is the reason why this software hasn't got a more descriptive name to this date.

  • Based on this initial image.c program, Fred Wobus wrote a full program to display all data output from the VAX analysis programs.

  • Gradually the FORTRAN programs were converted into C and incorporated into Image. Each of the analysis steps could now be called by the click of a button. This was the first official release Image 1.0

  • More and more features were added to Image1.0 and the file formats changed. The first usable version of Image was Image 2.0, which is still used in a few places.

    Versions 2.1 to 2.3 contained bug-fixes and new features as fpc slowly came along, which is the UNIX replacement for Contig9.

  • In late summer 1995, Fred went off to rewrite all the Image code. The window layout was changed and cleaned up, and for the first time new analysis functions were tried out.

    This was a first prototype version 3.0 beta, which was never released, because it didn't work the whole way from scanning to output to the database.

  • Spring 1996:

    A new start was made by Fred Wobus and Darren Platt.

    Darren converted the new analysis functions from the beta prototype into C++ and made them into command line programs.

    This new version introduced the new layout and the concept of a processing pipeline that the user steps through be clicking forward and backward buttons. This idea was inspired by the Wizard approach of installation programs on Windows-PCs, because they provide good guidance to the user and only show data relevant to the current stage.

    Image 3 made a huge improvement on data throughput by cutting back on manual editing time.

    Fred left the Sanger Institute when Image3 was in a stable and highly usable state. The support was left to Darren Platt who continued to work on the analysis algorithms.

  • As fingerprinting was done on a bigger and bigger scale new features were implemented by Darren Platt to facilitate fluorescent fingerprinting. The ABI sequencers were used load multi-dye gels, which has samples loaded in each dye.

    When Darren left the Sanger Institute, Image was left running pretty much as it was - the software was stable and producing good results.

    Later John Attwood and Jim Mullikin of the production software team at the Sanger Institute took up the support for Image and in collaboration Image3 was maintained and improved. This resulted in the eventual release Image3.9, which was used very successfully at the Sanger Institute and St. Louis and around 40 other genome centres and laboratories for medium to large scale fingerprinting and mapping.

    Jim Mullikin did a lot of work on the development of the analysis algorithms. They were simplified, speeded up and tailored to the specific needs of large scale fingerprinting. He also adapted the "pathfinder" lane tracker for Image, where it was used to lane-track the ABI gels.

  • Another breakthrough was made when Fred Wobus and Simon Kelley teamed up again at the Sanger Institute to work on the graphics that was shared between Image and the Acedb software.

    Simon had converted the acedb graphics library to run on top of the GTK widget set. Image was going to be a showcase to demonstrate how new user-interfaces can be constructed using GTK widgets while leaving original drawing code mainly untouched.

    It didn't take long to produce a prototype of Image3.9 that was compiled against the GTK libraries. However this version would have looked exactly the same as before, and Fred was itching to produce good-looking user-interface which would look and feel like Netscape or a standard MS-Windows application.

    In the process, the entire program and every line of code was reviewed and reconsidered. In some cases the analysis programs had changed without adapting the interactive editor accordingly, in other cases external scripts and little gadgets were written to work around some of the limitations of Image. It was time to clean up the code.

    The result is Image3.10, which looks like a new program but still has all the functionality and much improved features as the trusty old workhorse Version 3.9. This time with a new sexy user-interface.

Last modified : January 2000 Image 3.10

* quick link - http://q.sanger.ac.uk/82ym4i2y