One of the things that I like about writing for Scientific Computing World is the freedom it allows me to deal with products because they are good, because they have something valuable to offer, rather than just taking what rolls in from the big publishers. Sometimes a suggestion comes from a reader, sometimes from the editor, a personal contact, a client, a chance discovery, a beta program. Sometimes, as here, I make a positive move to sample a new offering from a source with a proven track record. Biosoft is a small, dedicated software company in Cambridge (UK, not Cambridge Massachusetts - though they do have a US office too).
For more than 20 years, it has been producing effective, efficient, high quality, workmanlike tools without distracting frills, and at competitive prices. The main target market, as the company name suggests and a quick glance down the product range confirms, lies in the life sciences; but a significant part of its output is well worth investigation by anyone with general data analysis or visualisation needs. Unlike many these days, Biosoft doesn't produce new releases lightly, or as an easy revenue stream, so the appearance of FigSys, a replacement for its long established Fig.P graphics product, had to be worth a look.
- A composite from screenshots during a FigSys session. In the background is the raw data worksheet, with (at left) the 2D plot types pallette. Overlaying the raw data at midground are portions of the plot from an equation fit, a composite bar chart, and a 3D surface plot from continuous data series showing discrete behaviour. In the foreground are and extract from the steps in a graph wizard (centre frame), a data range selector (bottom right) and the rich 'MathType-like' (my own description, not that of Biosoft) properties editor.
Fig.P was written by an external author, starting life as a DOS program and subsequently moving to Windows. FigSys, by contrast, has been homegrown from scratch in the 32-bit environment, and is an evolutionary move rather than a simple replacement. Certainly, there is a clear strategy of carrying the existing Fig.P user base forward, but there are also gene-transplants from elsewhere in the stable - Stat-200 (Biosoft's generic statistical analysis product) and ModKine (a data modelling tool, also generic but with a particular slant towards pharmaceutical and other biomed users).
The result is a powerful and flexible, freestanding generator of illustrated technical documents. Director John Lamble describes FigSys as 'by far the most ambitious program ever written in-house', and examination of the result bears this out. Development spanned three years or so, and testing with Biosoft's trademark thoroughness has taken up most of another year. FigSys can quite happily do parts of the reporting process, working to other applications, if that's what you need, but it is really a complete page-oriented report and presentation package.
During the review period, I have produced material for every kind of purpose, from a printed paper report through a reference website to instructional slide show, all without hassle and all without using any other software. If you are used to Fig.P, Biosoft offers help in reproducing the results you are used to in the new setting; if you already know one of the other products in this same market, you'll soon feel at home. And if you are not used to anything like this at all, then let the wizards take over - there's one to fit almost every normal requirement, and they are an intelligent, intuitive bunch.
- Extracted from the help file, showing the interaction of components.
The operational structure is tight and well designed. An object-based workbook environment holds components (data sets, fits, plots, output layouts, wizards) in floating or dockable toolbars, and these are selected for the main pane. This workspace is highly configurable - so highly configurable, in fact, that I kept mislaying bits of it in the first rush of enthusiasm! Broadly speaking, before you start messing about with the default arrangements, it consists of the usual menu and toolbars at the top, an organisational pick list tree below it on the left with a tabbed choice of palettes below that again, a main panel to the right, and a log at the bottom.
As you work, the responsiveness is quickly apparent. I loaded the demo version onto an elderly 75MHz notebook with only 32Mbyte of RAM, and it still turned in a more sprightly performance than some well-known packages could manage on a 3GHz, 512Mbyte desktop. The model fits can be run free of constraints, or with any level of manually-set parameters, according to your requirements. Scratch the surface and you find Python, the scripting language used in Stat-200. The standard modules that make the application tick are all written in this, and you can add your own extension modules as well, for potentially infinite expansion and tailoring.
FigSys' features list is extensive, and freely available on the Biosoft website, so I won't waste word count on a detailed inventory here. Suffice to say that everything you see on that list works perfectly. On the one occasion when I hit a problem (of my own making, I fear) and sent an e-mail call for help, the response was immediate (and in the holiday week, too). Also available from the website is a downloadable demo version of the program, so you can try it for yourself as well. The real question, then, is: why should you take the time to do so? The answer to that depends on your particular needs, and on how highly you weigh the value of getting exactly the right solution to meet them.
In 'Sometimes, size doesn't matter', published in the April 2003 issue of Scientific Computing World, I suggested that bigger is not always better in software selection; that fitness for purpose is the primary criterion, and this often means 'small is beautiful'. That came home to me again, with renewed force, here. We are so used to harnessing large regimes to a wide variety of tasks that we often forget the power of purpose-designed tools. I can think of many settings where adoption of a specialised solution would save its own cost on the day of installation. If you are generating illustrated reports or presentations based on data modelling and visualisation, regardless of your context, then not to take a close look at what FigSys can do for you would be a serious mistake. If you are in a pharma or life sciences setting, even more so.
