Here, we collect thoughts and opinions on what the MFX flagship experiment(s) should be in the LCLS-II-HE era. We collect the timeline of discussions that we had at the end of this page and summarize what emerges from them by categorizing between things we all agree on (Consensus) and things we are debating (Sticking points).

What's a flagship experiment?


Consensus

  • MFX identity:
    • high-throughput time-resolved multimodal imaging+spectroscopy for Biology and Chemistry.
    • SPI belongs at CXI/TXI
  • high rep rate enables:
    • rapid crystallographic screening
    • collection of larger datasets => better SNR
    • increased temporal sampling
  • high energy:
    • Spectroscopy: Mo K-edge to study nitrogenases with LBNL
    • Anomalous diffraction opportunities for phasing/anomalous difference density calculation.
    • SFX: "As mentioned in a recent paper (https://journals.iucr.org/s/issues/2019/04/00/xh5054/index.html): crystals absorb less (may be less relevant to XFEL experiments), the Ewald sphere curvature is reduced (benefitting data completeness), and we can move the detector further away from the IP (making space for other devices and instruments)."
  • high flux:
    • EXAFS (an example of signal poor experiment that could find a niche at MFX) on PSII

Sticking points

  • "specialization vs flexibility": specialization is key?
  • "not-low" background at MFX means:
    • we can get a clean signal when performing the experiment in a helium environment. Have there been any systematic studies of the data quality collected on the sample in vacuum vs helium? Are all sample delivery methods compatible with a helium environment (MESH, electrospray)? What do we know about helium availability and price in the future?
  • higher flux: "will this be an increase in the overall photons per second, or will we also have an increase in intensity per pulse? In case of the latter, we will not see an increase in signal."

Potential Collaborators

  • Lois Pollack: time-resolved solution scattering
  • David Baker: protein-encapsulated nanoparticles
  • Nate Hohman: SFX of small molecules


Timeline of discussions

Instrument meeting 6/27/23

During this meeting, Leland asked the MFX team to start brainstorming about identifying LCLS-II-HE flagship experiments starting circa 2026, highlighting the 3 characteristics that these experiments could benefit from, namely 1) High repetition rate (up to 1MHz), 2) High flux (more info?) and 3) High energy (up to 20 keV).

  1. High repetition rate
    1. Dan R pointed out that Single Particle Imaging (SPI) could benefit from it, to compensate for ultra-low hit statistics. In terms of samples, gold dumbells or protein-encapsulated nanoparticles (work from the Baker Lab; maybe see this recent preprint?) were mentioned. It was pointed out that the main challenge for SPI remains injection: Mark H mentioned ongoing work in that direction: maybe worth following up? Who is working on this, what is the current plan, etc.
    2. Fluctuation X-ray Scattering (FXS) would also benefit from high rep rate
    3. High-throughput SFX for Biology and Chemistry (see Nate Hohman). See ideas submitted to BRaVE3 for automated experiment steering.
  2. High flux
    1. Extended X-ray Absorption Fine Structure (EXAFS) would benefit from higher flux and could be used to study PSII with LBNL.
  3. High energy
    1. Again for spectroscopy, reaching Mo K-edge would help to study nitrogenases with LBNL (Mo pterin?)

Follow-up emails on 6/28/23

Following up, Fred P wrote:

Either way, I was thinking about this flagship experiment thing, and I guess what this is really about is defining the “identity” of the instrument. And I remember that someone (Mark? Ray?) once introduced MFX to me as “the beamline that produces a lot of SFX data/papers”. Another way that I think of MFX is by opposition to CXI in terms of background levels. So for me it would be natural to think of MFX as the SFX/solution-scattering beamline going forward, going after high-throughput and productivity more than it would go after challenging signals buried in the background or “one-off” experiments.

It could be defined as the beamline that delivers on high-throughput time-resolved multimodal bioimaging+spectroscopy experiments?

Another idea that might be behind defining flagship experiments is to reduce overlap between instruments. For example, would it make sense to push for SPI at MFX if it can be done more successfully at CXI or TXI? And likewise, do we really need to do SFX at CXI or other instruments if MFX does it so well? For solution scattering, the jury is still out: Lois Pollack would love to do it at MFX because it’s much simpler but in her current experience was only able to get good data at CXI, etc.

To which Leland answered:

These are excellent points you make. I struggle with the "specialization vs. flexibility" going into the future.  My impression from the LCLS-II HE retreat was that specialization is key going forward and spectroscopy+bioimaging seemed to be accepted as MFX's strong niche - so your point about SPI at MFX vs. CXI/TXI is reasonable. I feel some​ signal poor experiments could find a niche at MFX, for example bio-EXAFS experiments started a boom in the synchrotron world that we have not yet paralleled.

Sandra's email from July 7

- High repetition rate: this will this be useful for rapid crystallographic screening, but I am also very excited about increasing the temporal sampling within a single experiment. In time-resolved crystallography, we often collect only a few timepoints and have to deal with a mixture of active state intermediates. Better temporal sampling will allow us to see the increase and decrease in occupancy of these active state intermediates and may allow us to better deconvolute intermediates in the electron density.

- Higher repetition rates will allow us to collect larger datasets and push the SNR. This is very important for ultrafast experiments, in which the occupancy of the active state is very limited (due to e.g. short pulse duration of the femtosecond pump laser).

- Regarding CXI vs MFX, we can get a clean signal when performing the experiment in a helium environment. Have there been any systematic studies of the data quality collected on the sample in vacuum vs helium? Are all sample delivery methods compatible with a helium environment (MESH, electrospray)? What do we know about helium availability and price in the future?
- Regarding the increase in flux: will this be an increase in the overall photons per second, or will we also have an increase in intensity per pulse? In case of the latter, we will not see an increase in signal.

- Regarding the K-edge and X-ray spectroscopy: X-ray absorption is not just great for spectroscopy, it can be very informative in diffraction experiments as well. Absorption allows us to calculate anomalous difference density maps using MR-SAD, to localize the ion in the electron density and calculate its occupancy.

- An increase in X-ray energy is something that is also being explored at synchrotrons. As mentioned in a recent paper (https://journals.iucr.org/s/issues/2019/04/00/xh5054/index.html): crystals absorb less (may be less relevant to XFEL experiments), the Ewald sphere curvature is reduced (benefitting data completeness), and we can move the detector further away from the IP (making space for other devices and instruments).
- I am excited about the multimodal scattering-spectroscopy beamline idea. XRD+XES was also one of the highlights of the conference I recently attended, many scientists in the bio community are very eager to complement their time-resolved diffraction experiments.

Notes from July 13 meeting

(I'm still in the process of compiling... below is rough draft)

Sandra - points out confusion about MFX 2.0, MFX HE, BRaVE, etc.

Sandra - Does MFX HE aim for something like BRaVE or something fancier (attosecond, etc...) more flexible

Dan - Build a stable setup that enables high-throughput through automation, but leave the possibility initially to pull it out so time can be devoted to cutting-edge user experiment: balance stable with flexible initially. Eventually, integrate R&D in stable setup which will prevail over time.

Automation creds at Dan's previous beamline: video of a robot taking a sample from a 96 will plate, moving it to the IP, triggering data collection and analysis, stop when complete, wash and repeat.

Limiting factors of high-throughput:

  • Detector readout rate
    • expect 25 kHz?
  • Sample volume
  • Sample delivery rates:
    • DoD at 5-10 kHz
    • DoT: hurdles with sticking on tape.
    • liquid jets (demo at FLASH: GDVN at 4.5 MHz). 
    • high viscosity extrusion at high rep rate: need analysis to show feasibility

Ray - Yavas work assignment: MFX-HE will see the first HE light: high profile!

Ray - New KB: smaller focus for nano-crystallography would make MFX more appealing to traditionally CXI folks like Weik and Colletier

Ray - Pulse picker: which science cases would need it? Fixed target with ultrafast scanning needs PP.

Dan - Goal: zero/one operator system ==> complicated tape systems that require huge team would not fly. Liquid handling stuff is easier.

Ray - Nice-to-have: option to use tapes for world-class science

Tension between automation and world-leading science case:

  • Automation should free us so we can focus on more complex science cases
  • Automation means leaner budget
  • Automation means standardization which results in guarantees - something appealing to a wider audience, in particular industry

Dan's experience with automation: 4 years to build, last year they (Genentech) came.

Industry needs:

  • Todo: figure out the industry needs: Genentech, Insitro, Gilead, Altos, Amgen, etc...
  • They need a risk profile: what can they get if they pay for private beamtime; can they publish if they go for public beamtime
  • Ray's point about LCP and DMSO mixing issue (Viscosity challenge): impossible at synchrotrons, possible at XFEL, but viscosity is a major hurdle for high-throughput

Timeline: what's feasible in 1/3/5 years in terms of HTP system.

Our strong suite: 

  • room temperature
  • radiation damage free
  • time-resolved
  • + spectroscopy

How does that define us versus competition with synchrotrons and cryoEM?

How many structures a day? (revisit this thoroughly)

  • conservative: 1/hour
  • optimistic: 1/2min
  • sample volume challenge.
  • appeal for molecular movies, salt screening, etc.
  • performance metric that we can guarantee to attract users.
  • fXS: similar metric?

Flagship: in 3 years, state of cryoEM - what's left for XFELs?

One-stop-shop for Bio strategy? Couple MFX with other capabilities? ...

Document: there is a useful untapped science in solving static structures. Make that statement emerge from need sourcing.

Possible renaissance in comp.meth.dvlpmt for crystallography and ML: more info extracted from the same data. Opportunity for HTP

Flagship is the technique:

  • Goal #1 "Le Mans": from collection to structure in 1hour, tirelessly for 24 hours with 1 operator
  • Goal #2: from that baseline, go to time-domain: automatic sampling of reaction coordinates.

TODO:

  • Calculation: what can be delivered - reach out to experts
  • Figure out industry needs

11/7/2023

Questions from Leland:

  • What is the desired maximum energy MFX-HE should be capable of attaining?
    • What experiment are we trying to accommodate at the very high energies?





  • No labels