Doodle poll: flow cytometry meeting

30 views
Skip to first unread message

Jake Beal

unread,
Nov 22, 2016, 12:04:57 PM11/22/16
to sbsc-flow...@googlegroups.com
As discussed in our last meeting, we're going to have a strategic planning session with Mark and Sarah next week.  I've created a Doodle poll to determine the best time for this meeting.  If you are interested in attending, please fill it out:


Thanks,
-Jake

jakebeal

unread,
Nov 23, 2016, 1:41:57 PM11/23/16
to SBSC Flow Cytometry Working Group
Looks like the best time for everybody who has responded is Friday afternoon.
I have chosen noon Pacific / 2pm Central / 3pm East Coast.

Google hangouts link for the call:

Thanks,
-Jake

jakebeal

unread,
Dec 1, 2016, 3:50:53 PM12/1/16
to SBSC Flow Cytometry Working Group
Here are slides to seed our review and discussion at the meeting tomorrow.

Thanks,
-Jake
2016-Q4-FlowStandardization.pptx

Brian Teague

unread,
Dec 2, 2016, 10:48:53 AM12/2/16
to sbsc-flow...@googlegroups.com
These look to-the-point, and I think they'll support the conversation we
want to have.

Brian
> <http://doodle.com/poll/a9zsqb3c2hbsz5nk>
>
> Thanks,
> -Jake
>
> --
> You received this message because you are subscribed to the Google
> Groups "SBSC Flow Cytometry Working Group" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sbsc-flow-cytom...@googlegroups.com
> <mailto:sbsc-flow-cytom...@googlegroups.com>.
> To post to this group, send email to
> sbsc-flow...@googlegroups.com
> <mailto:sbsc-flow...@googlegroups.com>.
> Visit this group at https://groups.google.com/group/sbsc-flow-cytometry.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sbsc-flow-cytometry/deb4a77a-81a1-402a-99c4-0e5cfacfc50d%40googlegroups.com
> <https://groups.google.com/d/msgid/sbsc-flow-cytometry/deb4a77a-81a1-402a-99c4-0e5cfacfc50d%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.


--
Brian Teague
tea...@mit.edu
Weiss Group, Synthetic Biology Center @ MIT

Marc Salit

unread,
Dec 2, 2016, 11:04:27 AM12/2/16
to sbsc-flow...@googlegroups.com
Thanks Brian and Jake --

One thing I wish for -- some examples of where upgraded flow cytometry data would have been (amazingly) valuable.

Are there a couple or few papers where comparability of scale across conditions/channels (or across different instruments or labs) would have made the paper more useful, or made it so results could be carried forward in a way that they can't without it?

It'd be great to have some papers like that to talk about.

In other words, what do you really wish for? I haven't seen a profound scientific value of calibrated flow cytometry for it's own sake. All I have is a leap of faith that comparability would change and improve practice. Do you have examples of "if only we had comparability, then..." ???

Thanks!
Marc
> To unsubscribe from this group and stop receiving emails from it, send an email to sbsc-flow-cytom...@googlegroups.com.
> To post to this group, send email to sbsc-flow...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sbsc-flow-cytometry.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sbsc-flow-cytometry/f8a668d7-dcd8-0770-cf27-f3c36ec3cfce%40mit.edu.

Marc Salit

unread,
Dec 2, 2016, 11:30:40 AM12/2/16
to sbsc-flow...@googlegroups.com
And, by the way -- if it's helpful to frame the discussion, I think of the most typical use scenario as this: the measurand in flow cytometry is enrichment of a phenotype in a population of cells.

Maybe we can enumerate a portfolio of measurands, and the scenarios they correspond to?

  • I'm trying to identify coincidence of two phenotypes in a cell (we usually probably do this with a general expression control in one channel and a control of interest (a reporter for our circuit of interest) in a second channel. Measurand here is the population density and location of the coincident events...

If we can identify a suite of measurands, then we might be able to better describe the value of establishing a "calibration" scheme for each.

Thanks,
Marc

jakebeal

unread,
Dec 2, 2016, 1:07:12 PM12/2/16
to SBSC Flow Cytometry Working Group
Here’s some examples where I’ve needed and didn’t have comparable units, and thus had to abandon attempts to build on these otherwise excellent publications:

- Mutalik et al., “Precise and reliable gene expression via standard transcription
  and translation initiation elements,” Nature Methods, 2013.

  The a.u. means these can’t be imported into any model-driven design programs.
  When we’ve wanted to use these, we’ve had to re-characterize from scratch, and
  thus limited our use greatly.  We also had real problems knowing whether the
  problems we ran into with BCDs were inherent to the system or problems with
  our use.

- Bonnet at al., “Amplifying Genetic Logic Gates,” Science, 2013.

  These gates are really promising, but have some performance problems that 
  limit their application and the complexity of circuits that can be built with them.
  If this publication used calibrated flow, we could apply the models developed
  elsewhere to figure out which of several possible problems is dominant, and 
  thus which approaches to improvement are likely to be most useful, which
  could allow us to build really high-performance sense-and-control systems.

- Nielsen et al., “Genetic circuit design automation,” Science, 2016.

  The relative units in here mean that anybody who’s not in Chris Voigt's
  lab is likely to have a real problem getting these circuits to run
  in their lab.  It would also be a lot easier to figure out what’s going wrong
  in the systems they aren’t able to predict well or that misbehave if we 
  knew the relationship to the typical range of absolute expression levels.


I’m also involved in a project starting shortly where we’re applying calibrated flow to get a better handle on engineered CAR-T specificity and performance, where quality control and predictable behavior is currently a serious obstacle to effective transfer into clinical application.


Also, here’s some examples where comparable units have already let us do really cool things:

Prediction of repressor cascades and feed-forward circuits from models of individual repressors, to better than 2-fold precision:

  Noah Davidsohn, Jacob Beal, Samira Kiani, Aaron Adler, Fusun Yaman, Yinqing Li, 
  Zhen Xie, and Ron Weiss, Accurate Predictions of Genetic Circuit Behavior from 
  Part Characterization and Modular Composition, ACS Synthetic Biology, 
  doi: dx.doi.org/10.1021/sb500263b, 4 (6), pp 673-681, June 2015.

Prediction and control of expression levels of RNA replicon mixtures to better than 2-fold precision:

  Jacob Beal, Tyler E. Wagner, Tasuku Kitada, Odisse Azizgolshani, Jordan Moberg 
  Parker, Douglas Densmore, and Ron Weiss, Model-Driven Engineering of Gene Expression
  from RNA Replicons, ACS Synthetic Biology, dx.doi.org/10.1021/sb500173f, 4 (1), 
  pp 48--56, January 2015 (supplementary information).


Development of CR-U6 devices (we would only have had the less effective CRPs without calibrated flow comparisons):

  Samira Kiani, Jacob Beal, Mohammad R Ebrahimkhani, Jin Huh, Richard N Hall, 
  Zhen Xie, Yinqing Li and Ron Weiss. CRISPR transcriptional repression devices and 
  layered circuits in mammalian cells, Nature Methods, 11 (7), pp. 723-726, July 2014.

Thanks,
-Jake


--
Brian Teague
tea...@mit.edu
Weiss Group, Synthetic Biology Center @ MIT

--
You received this message because you are subscribed to the Google Groups "SBSC Flow Cytometry Working Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbsc-flow-cytometry+unsub...@googlegroups.com.

John Sexton

unread,
Dec 2, 2016, 2:40:57 PM12/2/16
to sbsc-flow...@googlegroups.com
if it's helpful to frame the discussion, I think of the most typical use scenario as this: the measurand in flow cytometry is enrichment of a phenotype in a population of cells.

This is not how I describe my flow cytometry use case. I use flow cytometry as a proxy for transcription rate (via several major assumptions, the merits of which are for a separate discussion), wherein it is important to capture the analog nature of transcription (via the "brightness" of a fluorescent reporter signal). My flow cytometry measurements are akin to a voltage or current in an electrical circuit (I have a pretty strong ECE bias... forgive me), where 5V is not the same as 10V. "enrichment of a phenotype" sounds very binary (i.e. % cells above a threshold), where 5V and 10V may both look the same (in that they are both above 0V). Please forgive me Marc if I've misunderstood your description of your scenario.


I (naively) think it may be difficult to finding specific mentions of the utility of a universal scale in the literature because it either (a) effectively undermines your own (non-universal) data and conclusions or (b) is tantamount to calling out your peers for poorly characterizing their published systems. I haven't looked closely, though.


I want to underscore some of Jake's points here:

- Mutalik et al., “Precise and reliable gene expression via standard transcription
  and translation initiation elements,” Nature Methods, 2013.
 
  The a.u. means these can’t be imported into any model-driven design programs.
  When we’ve wanted to use these, we’ve had to re-characterize from scratch, and
  thus limited our use greatly.  We also had real problems knowing whether the
  problems we ran into with BCDs were inherent to the system or problems with
  our use.

Strongly agree with Jake's stated need to re-characterize existing systems from scratch any time you want to import a new system into your lab.

In work currently unpublished, I've "imported" 5 different chemically inducible transcription systems, and the first step is always a costly, time-consuming (months) re-characterization, and in some cases re-optimization of each system. The re-characterization would be completely unnecessary if these systems had been previously characterized on a universal scale. Furthermore, no one else can use my data (when published) if my characterizations are not universal.


- Nielsen et al., “Genetic circuit design automation,” Science, 2016.
 
  The relative units in here mean that anybody who’s not in Chris Voigt's
  lab is likely to have a real problem getting these circuits to run
  in their lab.  It would also be a lot easier to figure out what’s going wrong
  in the systems they aren’t able to predict well or that misbehave if we 
  knew the relationship to the typical range of absolute expression levels.

I agree with this. The fundamental issue here is the composition of genetically encoded circuits. To do this, you must describe your circuits in internally consistent units so that the output of one circuit can be fed as the input to the next circuit. This work underscores the importance of reconciling logic gate output signals with downstream logic gate input signals, but, as Jake stated, they've only done so within the context of their lab. For me to pick up one of their circuits and tweak it in any meaningful way, I have to either (a) re-characerize their system with my instrument setup or (b) faithfully reproduce their instrument setup, which may be difficult or impossible. By enabling a universal scale, we make it much easier to effectively reproduce someone else's experimental setup (you run some calibration beads and use some software on your cytometer instead of buying a new cytometer).


A final point I wanted to make: a universal scale enables a constrained characterization (augmenting common unconstrained characterizations like the Fold Change metric). I made a silly little diagram to illustrated this (see attached), but I think this issue is still very important as the Fold Change metric represents an inherently incomplete metric.

-John Sexton


To post to this group, send email to sbsc-flow-cytometry@googlegroups.com.
augmenting_fold_change.png

Sebastián Castillo Hair

unread,
Dec 2, 2016, 2:52:11 PM12/2/16
to SBSC Flow Cytometry Working Group
I think one additional problem that universal calibration would solve is comparing part performance. In E. coli, there are lots of promoter libraries that either respond to combination of ligands (e.g. http://pubs.acs.org/doi/abs/10.1021/sb500262f), are controlled by transcription factors (http://www.nature.com/nchembio/journal/v10/n2/full/nchembio.1411.html, http://msb.embopress.org/content/3/1/145.long), or are simply constitutive, with different levels of composability (http://www.nature.com/nmeth/journal/v10/n4/abs/nmeth.2404.html, traditional Anderson iGem library).

We have had several cases in our lab in which we had to select some of these for multi-gene constructs. In this case, knowing how these promoter libraries compare to one another would have been tremendously useful. The alternative (which is what ended up happening) is to recharacterize these promoters in-house from either one library or more. This is time consuming (on the order of months), and tends to bias our next promoter selection towards the library that we have already characterized, independently of whether this is a good idea. It is also hard to convince a PI that spending time on comparing different parts from different libraries is worth doing, because the results would either be a supplemental figure on your paper (at best) or a paper that would be hard to publish in a high-impact journal, given that the numbers obtained are in a.u. and are therefore not useful to anybody else. The result is that several labs would have to repeat such characterization independently and spend a lot of additional time.


On Fri, Dec 2, 2016 at 12:07 PM jakebeal <jake...@gmail.com> wrote:

To post to this group, send email to


--
Brian Teague
tea...@mit.edu

Weiss Group, Synthetic Biology Center @ MIT

--
You received this message because you are subscribed to the Google Groups "SBSC Flow Cytometry Working Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbsc-flow-cytom...@googlegroups.com.
To post to this group, send email to sbsc-flow...@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "SBSC Flow Cytometry Working Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbsc-flow-cytom...@googlegroups.com.

To post to this group, send email to sbsc-flow...@googlegroups.com.
Visit this group at https://groups.google.com/group/sbsc-flow-cytometry.
To view this discussion on the web visit https://groups.google.com/d/msgid/sbsc-flow-cytometry/8c3f4567-97e4-4e3f-a87f-05b617276e56%40googlegroups.com.

Brian Teague

unread,
Dec 2, 2016, 2:58:37 PM12/2/16
to sbsc-flow...@googlegroups.com
On 12/02/2016 02:40 PM, John Sexton wrote:
> if it's helpful to frame the discussion, I think of the most typical
> use scenario as this: the measurand in flow cytometry is enrichment
> of a phenotype in a population of cells.
>
>
> This is not how I describe my flow cytometry use case.

+1.

I think of flow cytometry in synthetic biology as used to characterize
the distribution of cell state in a population, where that cell state is
represented by a fluorescence value (and is usually not binary!!). What
I generally want to see is how the parameters of that distribution
change as I change an experimental condition, e.g. the concentration of
an inducer.

Sebastián Castillo Hair

unread,
Dec 2, 2016, 9:16:08 PM12/2/16
to sbsc-flow...@googlegroups.com

First, thanks to Marc for meeting with us. We really appreciate your perspective on these issues.


I wanted to point out that there has been some characterization on how good the current methods (i.e. converting fluorescence to MEFL) are when using calibration beads for synthetic biology purposes. We have already demonstrated that:

Unfortunately, the current methods do not necessarily result in comparable numbers between different instruments. When the fluorophore of the calibration particles and the biological samples have matching emission spectra, calibration with these particles results in the same numbers on different instruments (see https://www.ncbi.nlm.nih.gov/pubmed/8809477, figure 1). However, this is often not the case in synthetic biology, where we try to calibrate fluorescent proteins to small-molecule dyes (e.g. GFP to FITC).


To show this experimentally, we measured the same 5 samples of E. coli cells expressing different levels of GFP in two different flow cytometers with different emission optics for the FITC channel, and calibrated them to the same calibration bead sample. Even after converting to MEFL, we observe a nearly two-fold difference between the fluorescence values of both instruments (see plot below, each dot represents one biological sample with MEFL fluorescence values as given by one instrument (x axis) or the other (y axis)).


pasted2


One way to solve this is using beads labeled with the appropriate fluorescent protein (e.g. GFP). To our knowledge, there is only one company that sells calibration beads based on one type of GFP (AcGFP) and one type of RFP (mCherry): http://www.clontech.com/US/Products/Fluorescent_Proteins_and_Reporters/Flow_Cytometer_Calibration_Beads/AcGFP_and_mCherry. People in the field use many more fluorescent proteins (we use superfolder GFP, some others use EGFP, YFP, CFP, etc).


Another way to solve this is to take into account the spectral differences between beads and fluorescent proteins, and compensate for them. This is what has sparked our recent interest in working with fluorophore spectra.


Hopefully this is helpful.


(Thanks to John for help editing this message)


--
You received this message because you are subscribed to the Google Groups "SBSC Flow Cytometry Working Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sbsc-flow-cytom...@googlegroups.com.

To post to this group, send email to sbsc-flow...@googlegroups.com.
Visit this group at https://groups.google.com/group/sbsc-flow-cytometry.
Reply all
Reply to author
Forward
0 new messages