PUMA for homemade arrays
Maitreya Dunham September
2007
Homemade arrays get split
and rotated while scanning.
Using SFTP, upload the red
and green tiff files for each array into your incoming directory on
loader. SFTP them onto the PC.
Open Genepix 5.1 in
Analysis Only mode. Emulate
scanner 4000B.
Click the Options button
(on right, image of little checkboxes).
In "File Open"
tab, under scaling, make sure pixel size is set to 5 µm.
Click the little disk icon
on the right and "load array list."
Select the .gal file that
goes with this print (can download from PUMA if needed).
Under the disk icon,
select "open images" and select the 2 images for an array.
Tell it which one is red
(635 nm) and green (532 nm).
The exclamation mark
button (on left) autoscales the image.
The gain and brightness sliders can also be adjusted.
Go to block mode (B).
Select all blocks by
dragging or with ctrl-A.
Position them roughly in
the center of the array. Use the
magnifying glass to zoom in if need be.
Under the spot target icon
(on left), select find array, find all blocks, find all features.
Use the greater than/less
than keys to cycle through all the blocks and make sure they're aligned.
If a block isn't aligned,
select the block and reposition it.
Hit "find features in selected block" under the spot target
icon (shortcut F5).
Once all the blocks are
aligned, switch to Feature mode (F).
Select all the features in
a block and hit L to cLear the not found flags (the ones with the bar through
the middle).
Move and/or resize any
misfound spots. To resize, hit
crtl and the arrow keys. If the
program has mis-sized a lot of spots, you can change the settings under the
Options button.
To flag as bad, hit A (for
Awful, I suppose). Flag obviously
bad spots like dust flecks, scratches, bubbles, etc. If a spot just doesn't look right, go ahead and flag
it. The most important thing is to
decide on your standards and stick to them.
Under the disk icon,
select "save settings" to save a .gps file with the grid settings.
Once you're done flagging,
hit the Analyze button (on right, looks like a table of numbers).
A big table of numbers
will open. Save this under the
disk icon with "save results" to get a .gpr file.
FTP the .gps and .gpr files to loader.
Log in to the
database. Under Enter Data, click
Experiments and Results.
Click Enter a New
Experiment into the Database.
In the database, select
Spotted as array type, Genepix as analysis software, and Saccharomyces
cerevisiae from the 3 pulldown menus.
Click Enter a New Experiment.
Fill out the experiment
description form. Choose the print
from the menu. Make sure all the
files match. Pick a logical
category and subcategory. If the
reference was in the red channel, click the dye swap box. Select your collaborators from the list
to give them access.
Put in as much information
as possible. You will probably not
go back and add in extra documentation later. Do it now.
Submit the form to the
database.
You will get an email
confirming that the files have been submitted to the database, then another one
when they have been actually placed in the database. Go to the URL in the second email and look at the submission
data. Make sure everything looks
OK.
To look at the data, go to
the search section, select any subcategory or user filters, and hit data
retrieval and analysis.
Select the experiments you
want to analyze from the list.
Choose Data Retrieval and
Analysis.
I typically retrieve data
by SUID, which averages all the spots that are for the same gene, as long as
they pass your filters. On
occasion, you may want to retrieve by spot to make sure your replicates are
actually correlating.
Select any biological
annotations you want.
Choose what label you want
on each experiment.
Proceed to Data Filtering.
I usually retrieve the Log
(base2)(REDsignal/GREENsignal).
Invert any dye swaps.
Choose your filtering
criteria. I usually use
2.5X>background in either channel (i.e, 2 OR 3).
If you would like to
actually inspect the spots, click the box next to retrieve spot coordinates.
Download the resulting .pcl file, or proceed to filtering/clustering as desired. Or add it to your repository.