Dictating DNA sequences to a computer

hey at mbcl.rutgers.edu hey at mbcl.rutgers.edu
Tue Nov 22 10:46:29 EST 1994


Dictating DNA sequences.

   I thought others might be interested in the way I have recently been 
entering DNA sequence data, from autoradiograms, to a computer.  The system is 
easier than using a digitizer, and comprably priced if you already own a quick 
PC clone (probably cheaper, I have not recently priced audio digitizers). 

The components:

Quick PC clone microcomputer (I use a 486 66Mhz Dell) with at least 16 meg 
RAM.

IBM OS/2 2.1 or 3.0  operating system.

IBM Personal Dication System for OS/2:
   This includes a hardware component, (a plug in card), a 
   microphone, and a software component.  The hardware and software sell 
   separately and both are about $500. I bought mine thru Indelible Blue 
   (800-776-8284). Contact IBM, not me, for more info.

Rimstar programmer's editor - OS/2 version. 
   Rimstar Technology Inc. (603) 778-2500.  This is also available thru 
   Indelible Blue or directly from the company.


How it works:

   The dicatation system works in two modes, dicatation and application 
control.  The dictation mode is nifty and very impressive, with a 30,000 word 
vocabulary that can be augmented. However in my experience it is not ideal for 
entering DNA sequence data.  Instead I use the application control feature of 
the dictation system.  This is basically a system whereby navigation through 
applications and windows that are open within the graphical interface of OS/2 
is done by dicatation of commands.  Under application control, the dictation 
software only recognizes a relatively small set of words, i.e. those that 
currently make sense for the windows that are open.  It is also quite easy to 
dictate macros to add additional functions to application control.

   It turns out that the Rimstar editor can be directed quite well under 
application control dictation. Furthermore this editor is highly configurable, 
and has a full featured macro language (very much like C) with a built in 
compiler.  What I have done is to create a set of dictation macros so 
that I can speak a small set of commands that cause the editor to write the 
different bases of a DNA sequence.  Each dictation macro triggers a macro 
function written in the Rimstar macro language.
   The different commands I use are associated with the following functions:
   - write 'A' and beep at  800 hz for 150 millisecs.
   - write 'C' and beep at 1200 hz for 150 millisecs.
   - write 'G' and beep at 1500 hz for 150 millisecs.
   - write 'T' and beep at 1800 hz for 150 millisecs.
   - write 'N' and beep at 400 hz for 150 millisecs.
   - delete the previous character and beep twice.
   
   The beeping provides an audio cue that the editor has received the 
intended command. All other Rimstar menu commands are also accessible by 
dictation. In particular "save" is frequently used to save the file.

   Any choice of words can be used for these functions, including the simple 
words "A", "C" etc.  However the dictation system does require clear speech 
with at least very short pauses between spoken commands. In practice I have 
found a different set of words to be more forgiving of vagaries in speech and 
less likely to be mistaken as other Rimstar menu commands. I use 
   - "ALL" for the A function.
   - "CHEESE" for the C function.
   - "GRAIN" for the G function.
   - "TICK" for the T function.
   - "N" for the N function.
   - "BACKSPACE" for the delete function.

   Probably the biggest downside to all this is that you sound a little goofy.

   By the way, Rimstar also makes a great sequence editor, especially for 
multiple sequences.  It has essentially unlimited line lengths, very easy 
column blocking and movement, and is highly configureable.  Brian Smith, the 
author has even written a macro that causes the four letters A, C, G and T to 
appear in different colors. Very handy for allignements.

Jody Hey
Department Biological Sciences
Rutgers University
hey at mbcl.rutgers.edu


   



More information about the Methods mailing list