General issues/wish-list

[lars.b...@gmail.com]

Hello Opera users.

We all need to know where PE is going with the Opera for the future or
at least the next 2-3 years. It would be great if anyone has any info
they can share or a way to extract information from PE, verifiable and
trustworthy information, regarding what is planned for the next 2 -3
years.

The release of the Operetta was, in my opinion, a step in the wrong
direction for users of the Opera since there are products of that
nature already on the market thus indicating a dire need for solid
advice from actual high-end system users.

Here's a list of features I would like to see included in the next
version of the Opera, I will call it "Amadeus" in the text and I will
include software features into this. Please add to list and re-post
it.

Amadeus features:

1.
AOTF for ultra-fast simultaneous image aquisition with 2 or more laser
lines against 2 or more cameras. By including this, the light-path
could be optimized in a rather dramatic manner thus minimizing light
loss from dichroic's and other mirrors as well as from sheer distance
loss.

2.
Plug-and-play camera positions if the AOTF is implemented. I would
like to see up to 8 positions available not only for the multiple
wavelengths but also for the different types of cameras that could
then be used for super imposing images (sensitive time-laps with a
multi-colored high-res every 10-20th frame would enable so much more
information extraction)

3.
Caged optic path. Seal the optic path to avoid particle contamination.

4.
Stage to be re-designed. The stage needs to be more forgiving or at
least able to self-adjust for plate re-positioning.

5.
Script writing module for object oriented programming. The script
could be outlined in a more graphic interphase and, if needed, fine
tuned later. Would speed up the writing of more complex scripts.

6.
Multi-laser diode bays. Ability to add multiple laser diodes for each
line thus increasing the energy for each line. Up to 4 positions.

7.
Enhanced liquid handling ability as an option (enhanced compared to
the option existing today). ( Wet Amadeus)
A. Implement a non-contact liquid dispensing capabilities to the
system in conjuncture with:
B. A user-loadable cartridge system for reagents thus enabling a
larger array of reagents being available, avoiding disposable tips,
cross contamination due to contact delivery.
C. Protective layer dispensation after reagent addition, see
below.


8.
An overall design enabling stable plug-and-play features to be added:
Second scan-head with different sized apertures; other types of
(confocal/non-confocal) microscopy units to be added.



Non-technical feature:

A user-group work-group with direct communication to the R&D team at
PE developing the Amadeus. Why not instate a formal Amadeus
development consortium between all present users. I realize that there
are an mountain of red-tape for this to happen but that would
potentially give us a product with capabilities and design features
far better than anything on the market today or that will come onto
the market for a rather long time.

I also recommend anyone using the environmental chamber to overlay the
wells with ESC certified mineral oil to negate the water-loss and also
to use pH buffered media. This should in a 37C environment give you at
least 6 hours of reactive, live cell imaging without the issues of
stable 5%CO2 and 100% humidity, 37C is enough. Use a dedicated multi-
drop for the addition prior to imaging.

[lars.b...@gmail.com]

Hi again,

Just wanted to clarify and add to my comment on the light-path:

A decrease in the light path length from 35cm to 20cm would give an
increase in energy from 100mW to 306mW due to :

E2=((d1/d2)E2*E1))

Due to the fact that I actually peeked into some parts of the Opera
(yes, I am very curious) I feel that some input from the global Opera
user group just has to be able to trim a significant part of the light
path....especially if we could step away from the dichroics.

On another note, has any one actually seen how some of the dichroics,
mirrors and filters, especially how the filters are mounted. The
filter-mounting I was privy to inspect had some sort of "homemade"
contraption that positioned the filters at the right way, I will see
if I can take a picture of the thing and get your feedback on it. My
take on how that single piece was assembled is that if one thing is
assembled and designed that poorly I will be less than surprised if I
find more of the same caliber thus I would like to see if there is any
chance to convince PE to enlist the global user group as their "design-
team", or at least, as an advisor to avoid cataclysmic failure for
their high end HT imaging division...


On May 24, 7:25 am, "lars.bran...@yale.edu" <lars.bran...@gmail.com>
wrote:

[Emmanuel Gustin]

Hi Lars,

I thought I would help feeding the board, as it could be a very useful
resource to all of us. I don't think PE is going to share their
business plans for the next years with us, but most suppliers are
willing to listen to user input...

1. Could you explain in more detail what you suggest, because I don't
really see how an AOTF would fit in. Because it works on Bragg
diffraction, an AOTF works well to couple monochromatic excitation
light into a beam path, but it doesn't seem very useful to manipulate
the broadband fluorescent light, at least not if you want to retain
reasonable bandwidth and a sharp image? I also see some problems
matching the aperture of the AOTF to the aperture of the back of the
objective. To me it seems difficult to avoid the use of dichroics to
divide the fluorescent light over multiple cameras. You might in
theory use an AOTF to couple the excitation light into the beam path,
but the layout of the Yokogawa CSU confocal module assumes that the
excitation light passes through both microlens and pinhole disk, and
the fluorescence is coupled out by a beamsplitter between the two
disks.

2. I can see why some academic labs would love this, but I doubt most
users would extract much benefit from it, and it has additional cost.
The problem is not just the hardware, but also the software. I think
this is the kind of feature that most people would regard as an
advantage when ordering the instrument, but never actually use.

3. I agree that the more the optical path is protected the better,
although a perfect seal is probably impossible to achieve.

4. Yes, the stage is a weak point of many HCS instruments and
automated microscopes; I don't really understand why so many people
continue to rely on spindle stages driven by stepper motors without
feedback... except that these are "relatively" cheap. AFAIK the
Operetta has a better stage, with higher precision, and I hope that
they will fit it also to the Opera.

5. I am a bit worried about the future of Acapella. The problem is
that to use it effectively, user groups need software and data
analysis skills that many biology teams don't have on board. In this
century of high content screening, phenotypical analysis, systems
biology, and -omics, cell biology teams probably need to possess such
skills anyway, but the acceptance of this is slow and far from
general. Yes, a cleanup of the script language to use more modern
software concepts would be desirable, but the benefit of this is tied
to the evolution of the user community.

6. I'm more inclined to bet on the appearance of more powerful lasers.
Aligning the fiber coupling is hard enough as it is.

7. Would be nice, although again I am not sure how many people would
really use it. There are better options than the traditional
injectors, but injecting liquids on the system is only advantageous if
you do fast kinetics. If not, then use a benchtop dispenser and save
yourself the worry of liquid spills on the optics.

8. I think this is reaching the point where "build vs. buy" comes in.
If you want that kind of flexibility, and you have the skills to
actually use it, you are probably better off starting with an optical
table, a dark room and a basic optical kit. Most of the components you
need are individually available, although you would need to have the
right skills to design the beam path of a microscope, which isn't as
simple as it may appear.

Best Regards,

Emmanuel

[gbo...@gnf.org]

Lars,
Thanks for starting such a fruitful discussion. I fully agree with
you, Evotec/P&E, should give their customers some idea about what
their vision of the future holds for the Opera, but I doubt that P&E
would share more than a broad vision of their plans.
I agree that the Operata was a mistake, because this new instrument
does not have anything special compared to that of their competitors,
so all they are doing is cut another slice of a small pie, rather than
keep their domination for very high-throughput microcopy. That being
said, I am sure they have done some market analysis and decided that
another imager would be a good investment. Unfortunately for us (ie
Opera users), this new development was done to the detriment of the
Opera. Instead of having a new team dedicated to the operetta, they
devoted all of their efforts on the Operata, and sacrificed fixing/
improving the current Opera platform (this might have been painfully
obvious for those of you who ran into problems with their instruments/
software last year).

Before talking of the future replacement of the Opera I think some
small revision of the current platform are in order. First I think
that many improvements on the software side would tremendously improve
the entire platform. Here are some improvements that I believe should
be made to the current system (and perhaps could be used in future
versions of the hardware).
1) Automate the CCD calibration/registration, and improve the quality
of these corrections with multiple acquisitions/averaging.
2) Automate plate registration and stage registration.
3) Take into consideration the angle of the stage, so that the center
of each well is correctly calculated. This is particularly important
when working with small wells. Right now only the top-left and the
bottom-right wells can be perfectly centered.
3) Provide a new script development interface. I would recommend
developing an eclipse plugin for Acapella. I would also encourage them
to look into KNIME/HCDC to allow users to do follow up data analysis.
This could integrate nicely with the Columbus solution.
4) The distribution of the image analysis jobs, should not be fixed
and should be assigned on the fly. Perhaps this could be done as part
of request #3.
5) There should be tools to restart an unfinished acquisition and to
image missing wells.
6) Large images should be streamed to disk rather than saved in
memory.
7) Improve memory consumption in general (cf. point6).
8) Provide a proper bug tracking site for their different piece of
software, with real time tracking of the tickets.
9) Provide a way to use the Opera as standalone microscope (ie, load
skewcrop/ref + Script + exposures-section) to get pictures of your
wells (particularly useful for assay development).
10) New and improved calibration plate. Perhaps using fluorescent
plastics etc.

For the future version of the hardware here are some of my wishes:
I believe Amadeus (ie Opera replacement) goals should be to increase
the throughput, so that a 1536-well plate can be read at high
resolution in ~5-10 minutes per plates (with let say 1 large field per
well). Currently the max throughput of the Opera is ~25” for 3
channelsx1 fieldx1Z-planesx1536-wells. Such increase will need to be
done through several strategies:

1) adding larger CCDS (perhaps more sensitive)
2) increasing laser power.
3) Future version of the system should use Z and x/y encoders to
improve the reproducibility of the stage and z-piezo positioning and
use a faster stage/piezo. The voice-coil technology with an optical
stage encoder such as the one used by the Meta-Express ultra is a good
example of what advantages this offers. Autofocus is one of the
slowest part of the process in the current version of the Opera (ex.
reading a 1536-well plate with 40ms exposure takes ~25" vs ~30" for
240ms exposure). Perhaps this means comming up with a new autofocuing
mechanism which could be measured a head of time (ie. separate optical
path).

Some other improvments are also in order:
The alignment/collimation of the CCDs should be improved. I think that
developing anisotropy-based FRET imaging would be a great option.

The stage loading should be done on the short side (rather than on the
front)! I think such a version of the Opera was developed for the guys
at Genentech.
For reagent dispensing, I don’t know if anyone really played with this
idea, but perhaps, we could have a system of cartridges like in a
IncJet printer to dispense low volumes of solutions much like an inc
printer does (perhaps this is what you meant in point#7B?).

Finally, I think your last comments about simple modifications which
may drastically improve the opera are spot on. Like you I agree that
there were some engineering flaws in the Opera (but this usually comes
hand in hand with new technology). I remain hopeful that with the
feedback of the community these errors will be fixed.
P&E remains in the lead for high-throughput cell-based imaging with
the Opera, although some new players are coming out (there is a new
HCI on the market from Yokogawa, but it does not seem to offer a
significantly better performance over the Opera and has a 2M price
tag: http://groups.google.com/group/highcontent/web/Yokogawa%2520CV6000.pdf).
To keep in the lead, I think addressing some of these issues in future
revision of the hardware/software is critical!

Best,

Ghislain

[Emmanuel Gustin]

Hi Ghislain,

Early this year, after being presented the Operetta, I asked a PE
representative why they had developed a direct competitor for other
HCS imagers. The answer was, basically, that this is a market segment
that is growing quite fast, while the sales of the Opera have remained
stable. It make sense that they needed to put an instrument in that
market. Whether it is wise to offer an instrument that offers no
obvious, large advantage over the competitors is something else: I too
would have aimed a bit higher, using the technology that PE already
has in house. However, they did make some improvements in the
Operetta, and I hope these will find their way to the Opera.

My own wishlist would overlap with yours and Lars'. Questions of
affordability, IP and practicality are not easy to answer.

For the instrument:

* An improved stage and focus. For an instrument with the price tag of
the Opera, we should consider a linear motor stage with accurate
optical encoders. That costs something, but it is far more robust,
very accurate, and I think you could save 100 to 200 ms per field on
motion and focus time. Vibration can probably be kept within limits if
the stage driver supports controlled speed/acceleration profiles.

* Improved illumination system, for greater long-term stability of the
light output and better illumination homogeneity. A traditional
problem with fiber-coupled laser illumination is its beam profile: The
illumination flatness is much inferior to that of a lamp coupled with
a liquid light guide. I wonder whether a liquid light guide could be
used with a laser system: It offers a larger entry diameter, better
efficiency, and a flat output profile. There must be ways to average
out the speckle (due to the coherent light source) during the exposure
time.

* Provide a mixed-mode autofocus that can use an image contrast
function as an alternative to, or in combination with, the reflection
autofocus. I know, this is going to be slower rather than faster, but
sometimes object-based autofocus works better.

* An additional light source under the hood for transmission images.
Some people have improvised this, and the Operetta has a LED light
source tuned to the wavelength of one of the emission filters.

* Better streamlining of the skewcrop and reference imaging procedures
to reduce the workload of these steps.

* For the injectors, the problem is the added complexity. Most
screeners still prefer the traditional peristaltic dispensers because
they are more rugged than inkjet or solenoid dispensers. Maybe the
microfluidics pump cartridges developed by Formulatrix would be an
option; it seems to be a compact and flexible piece of kit. But I've
never tested it.

* Better management of the experiment setup and metadata storage. The
current 'OperaDB' setup is very flexible, but it quickly gets rather
messy. I would prefer to have these data in a real database. Currently
we have to parse and transfer them to a real database ourselves.

For the image analysis:

* General cleanup of the Acapella script language. I think some of the
limitations are the result of their home-brew and old-fashioned
syntax. I admit that getting it redone the right way could be very
costly, but perhaps they could take an existing script language as
basis, instead of re-inventing the wheel. (Script languages such as
Tcl and Python were actually developed with that purpose in mind.)

* Have options to do cell-level reporting, preferably in a format that
conforms to the ISAC FCS3.0 standard for cytometry data. Using
Acapella, you can write and use your own reporting routines, but it
gets a bit tricky if you want to run those server side. Also, the
ability to store segmentation results as thin bitmaps (that can be
overlaid with the raw images) would be useful.

* During real-time analysis, have a command to trigger a re-visit of
wells, for example if there was a focus problem or the cell count is
too low. Then you can ask for a retry and/or the capture of additional
fields.

* Work with other suppliers and the user community on standards for
HCS images and data analysis results, and adopt those. That would give
all of us much more flexibility to mix and match our equipment and
data flows.

Best Regards,

Emmanuel