This menu allows you to generate a design for a two-colour microarray experiment.
A design can be generated by supplying the number of treatments and selecting one
of the three available standard types of design (Balanced Incomplete Block, Loop and
Common Reference Designs).
Types of Designs
Balanced Incomplete Block Design
A balanced incomplete block (BIB) design compares each treatment with every other
treatment an equal number of times. BIB designs are not automatically
balanced for dye, and the treatments are assigned randomly to
the two colours. Thus some swapping of the treatments within a slide may be
required to get the design balanced for the two dyes.
Loop Design
In these designs the treatments flow from one slide onto another with one
treatment moving to the next slide, but changing dye, and a new one being
introduced. When the treatments are exhausted, the treatment on the first
slide is put on the final slide to provide a loop back to the first slide.
The increment in moving between treatments is controlled by the Loop Increment
field. For three treatments, the loop and balanced incomplete block designs
are the same.
One property of this design is that they must be balanced on the two dyes
as each treatment occurs twice in each loop, one on each dye.
Common Reference Design
In these designs all treatments are compared once if a common reference
treatment. The reference treatment is specified in the Reference Level
edit field. A reference design is not normally the most efficient design
for a given number of slides.
Number of treatments
The number of targets to be used in the experiment. This must be at least
2 for a valid design. For a given type of design, the number of slides
that will be generated are shown below this field.
Loop increments
For a Loop design, the loop increment gives the steps between treatments.
The default loop increment of one steps sequentially through the treatments,
1,2,3,... . A loop increment of 2 with 5 treatments gives the order
1,3,5,2,4 (using clock arithmetic so that 5 + 2 = 2).
The loop increment must have no common factors with the number
of treatments, so for 6 treatments, increments of 2 and 3 cannot be used.
With 5 treatments, increments of 1,2,3 and 4 are all acceptable. For 9
treatments all increments a part from 3 and 6 are OK. Multiple loop increments
can be specified, in which case a design with that number of loops will
be generated. The increments in this case are given as a comma separated list.
This can be used to generate replicates loops with the same increment repeated
as in 1,1,1 which will give three replicates, or to generate distinct loops
as in 1,2,3.
Reference Levels
For a Reference design, the reference level gives the treatments that will
be compared with all other treatments. The default reference level is 1.
Multiple reference levels can be specified, in which case a design with that
number of replicates will be generated. The reference levels in this case are
given as a comma separated list. This can be used to generate replicates with
the same reference level repeated as in 1,1,1 which will give three replicates,
or to generate distinct replicates with a different standard in each replicate
as in 1,2,3.
Allocation of Reference to Colour
For a reference design, the reference treatment can be allocated to just
one colour (not normally recommended), or else randomized to colour or
else dye swap pairs can be produced. These options can be selected here:
| Red | - always allocate the reference treatment to red. |
| Green | - always allocate the reference treatment to green. |
| Randomized | - randomize the reference treatment to red or green. |
| Dye swap pairs | - produce two replicates of the design, with the reference
treatment allocated to red in on replicate and to green in the other. |
Randomization Seed
The randomization seed used to randomize the design. The default value of 0 uses
a seed taken from the clock and so will produce a different randomization
each time. A positive integer will generate the same randomization each time
it is used. Setting the randomization seed to -1 will generate an unrandomized
design.
Print Design
When selected, the design will be displayed in the output and for a Balanced
Incomplete Block design an analysis of variance (not allowing for dye effects) is displayed.
Spreadsheet
These options control whether the resulting design is displayed in a spreadsheet:
| Display Design |
Display the design in a spreadsheet. |
| Treatments in 2 columns by colour |
When selected, the spreadsheet will contain 3 columns,
Slide, Red and Green that have a
row for each slide, with the red treatment level in the Red column
and the green treatment level in the Green column.
Otherwise by default, the spreadsheet will contain 3 columns,
Slide, Colour and Treatment that has two
rows for each slide, with the colour each treatment is allocated to in the
Colour column and the treatment levels in the Treatment column.
|
Action Buttons
| Run |
Generate the design. |
| Cancel |
Close the menu without further changes.
|
Examples
Example Balanced Incomplete Block Design
This experiment compares every treatment of 1,2,3,4 with every other treatment.
The resulting design is displayed in the spreadsheet:
Note, the Balanced Incomplete Block design does not guarantee each treatment is balanced for dye,
as the treatments on each slide are allocated to dye at random. Switching the treatments between
the two dyes on some slides may improve the dye balance.
Example Loop Design
This experiment compares 5 treatments in two loops, one with increment
of 1 and the other with increment 2. A seed of -1 is used to generate
an unrandomized design so that the structure is more obvious.
The resulting design is displayed in the spreadsheet:
Loop designs are always balanced for dye as each treatment occurs
once on red and once on green in each loop.
Example Loop Design
This experiment compares 5 treatments all with the reference treatment 1.
Each comparison is done twice, in dye swap pairs, one with the reference on
red and one with the reference on green.
The resulting design is displayed in the spreadsheet:
See Also