This project involved creation of the control software for a
molecular biology workstation (MBW), which was a general purpose,
programmable chemical reaction controller. The workstation
supported a number of 96 well reaction plates, some of which
could be incubated under program control. The MBW featured
a flexible, robotic pipetting unit and was to allow the user
to create libraries of reaction protocols, which would permit
automated running of hundreds of reactions simultaneously.
The workstation design was flexible, supporting a wide variety
of different experiments including DNA sequencing. Among the
protocols supported were Sanger, Bst, Taq, Single-stranded
template, Double-stranded template, Fluorescent primers, Fluorescent
terminators, and Maxim-Gilbert.
The control software was to
support clear, efficient protocol creation and archiving
and a run-time executive that managed
reactions while permitting a vivid graphical display of the
current state of any of the hundreds of possible reactions under
at any given time.
The requirements definition for this project was extremely
well done and appeared stable from the outset. Therefore,
we decided to capitalize on this perceived stability by selecting
a staged delivery development methodology. This methodology
breaks the development process into milestones with the proviso
that each deliverable represents essentially the final form
of the modules in the delivered set. This methodology requires
that the project be partitioned into stages such that the
deliverables for stage N will work correctly without those
for stage N+1. The nature of this project was such that the
required factoring could be successfully done.
requirements definition was complete months before the
first hardware prototype was to be available from
the hardware contractor. Thus, the first software modules
to be designed and built were for the preparative operations
such as the protocol editor and archival management module.
Subsequent delays in the hardware prototype schedule forced
the development of an instrument simulator, which then
allowed construction, and testing of the actual instrument
and reaction display modules.
Thanks, in large part, to the excellent specifications for this
product, the software development effort was successfully concluded
after 13 months. Although this was technically one month beyond
the projected schedule, the software contained all of its originally
specified functionality plus a protocol optimizer, which was
added after the original schedule was created.
documents and internal embedded documentation allowed for further
development and maintenance by in-house developers.
Although we were on-call for problems in the field, we never
received any defect reports from customers. We were told
that users found
the GUI very intuitive and efficient to use.