Thread by @kirtiprakash25

Hey #STED, #MINFLUX people,
To continue the public discussion, I list here some of the points I made during my talk.

Summary: MINFLUX is a promising technology but at an early stage of development esp for biological research.

Thread👇

cc @Stefan_W_Hell @BalzarottiFran

Disclaimer 1: I haven't seen or worked with #MINFLUX setup, my arguments mainly on application-side

Disclaimer 2: I asked these questions via email (Sept 2020) and requested raw data and codes but haven't received a reply yet

I hope @rita_strack @naturemethods can help!

You can learn more about #MINFLUX from this excellent talk by @Stefan_W_Hell


Also, some papers making an effort in this direction

"MINFLUX imaging of nuclear pore complexes of a mammalian cell shows that this true nanometer-scale resolution is obtained in three dimensions and in two color channels"

Let's examine this claim but first what are nuclear pores and their dimensions?

Point 1: Molecular resolution or visualisation enhancement?

Please comment if I made any mistakes in my analysis, happy to be corrected!

Point 2:

1. Nuclear pores and true nanometer-scale resolution?
2. Is there a real need for the molecular resolution?

#MINFLUX

Point 3: Subjective filtering and molecular components?

Maybe it is all about the dots!

Please note that #MINFLUX does at least 2 levels of filtering - relative photon counts and background events.

Would be great to see the raw data before and after filtering @BalzarottiFran

Point 4: What happens to the molecular components with 2-color MINFLUX imaging?

While #SMLM methods (@JonasRies) can easily resolve both inner and outer ring of nuclear pores, #MINFLUX 👇

Deep thought: what's the goal of my research?

Point 5: The importance of apriori structural information

#MINFLUX has nicely resolved DNA origami at 5 nm, which has almost zero background when imaging. For NPCs, independent validation is still needed.

But what happens when you image proteins with no underlying structure?

Point 6: What happened to my microtubules?

For years, MTs have been the go-to structure for imaging

1. They have an "underlying structure" at 20-25 nm range
2. They are "dense", so any undersampling due to filtering will show up
3. They have been validated/studied so many times

Point 7: Molecular resolution of the fluorophore or protein of interest?

Not sure how to claim 1-3 nm resolution when linker itself is ~3 nm?

Also, not to confuse localisation precision with resolution!!!

Point 8: Raw data and 3D #MINFLUX

Again, where is the raw data?
And does the 3D data look a bit undersampled?

I fear people will ignore the points I make, so I tag most of the #microscopy people I know on Twitter.

And here is the rule of the game - you agree to the above points if chose to remain silent😃!

@HeintzmannLab @RetoPaul @AndrewGYork @JakobsLab @IlariaTesta4 @alesposito75

@kD3AN @koholleran @Sas_K_ia @EdwardAllgeyer @Fran_Bottanelli @Ulrike_Boehm @HanneJanina @Abberior @AsifaAkhtar1 @AICjanelia @EllenbergLab @HenriquesLab @seamus_holden @HohlbeinLab @YorkBioimaging @SteveTheChemist @michelledigman @KnerLab @MoernerLab @JLS_Lab @SjoerdStallinga

I just noticed that the original thread breaks here

Remaining thread with 8 points on #MINFLUX here

https://twitter.com/kirtiprakash25/status/1336579243313156101

Microscope users, facility managers,
What biological applications of #MINFLUX u can think of to convince the Institute head to invest in one?

@KRogers1001 @manorlaboratory @sossick @AlvaroCrevenna @StefanieReiche6 @koholleran @JenCWaters @YorkBioimaging @EllenbergLab @AICjanelia

I start.

I would try to resolve nucleosomes!
Unlike nuclear pores, it is a continuous structure and about 8-10 nm.

So would test the performance of #MINFLUX on biological structures < 10 nm.
Also, any excessive filtering would result in a discontinuous, undersampled structure.

Recording of my talk on #MINFLUX and super-resolution microscopy @RoyalMicroSoc



MINFLUX starts ~16 min

Loved the discussion session with @alesposito75 @StefanieReiche6 @sossick

Also, surprised to have more views in 1 day than Chris Dunsby in 3 weeks!

I have received >15 messages across different platforms last days briefly saying something like - "I like the points you make abt MINFLUX but I prefer to be on the safe side"

Fine if you want to be the silent spectator but pls don't forget the popcorn, more stuff to come😀😆!

Quiz on a Gedanken experiment!

What do you think is the first (< 22 nm) of the two peaks prominently seen in the localisation precision plot (2nd column)?

Dataset - GFP_AB-AF647_190517_1
ebi.ac.uk/biostudies/Bio…

Vote below!

Correct answer next week, but please get it right this time!

Now I discuss the questions asked by the participants of #ImagingONEWORLD


Q1: @alesposito75- #MINFLUX original paper (Balzarotti et al. 2017) does side by side comparison with other superresolution techniques?

No, PALM/STORM is only a simulation!

Here is how a side-by-side comparison should look like, according to me!

@alesposito75

And DNA origami is well-resolved by DNA-PAINT methods using a simple microscope setup at 5 nm range, see Strauss and Jungmann 2020
nature.com/articles/s4159…

In btw, Strauss/Jungmann got the #MINFLUX citation wrong - @rita_strack @naturemethods

The only citation of MINFLUX is - "visualized the nuclear pore complex protein Nup96 in U2OS cells expressing Nup96– mEGFP"

Gwosch 2020 didn't use Nup96-mEGFP, please correct me if wrong!

The first open-source implementation of #MINFLUX with lifetime info, ~4 years after the first publication!

https://twitter.com/l_masu/status/1339925149705859075

Congrats @l_masu @Fl_Steiner @PhilipTinnefeld @FerStefaniLab et al

Thanks for making the codes available.
And the raw data looks like 'raw' data😃!

@l_masu @Fl_Steiner, I notice that p-MINFLUX discarded events with <1200 photons (#MINFLUX with <500)

1. How do you decide on the threshold?
2. Does the prior knowledge of the DNA origami structure influence the threshold?
3. Is it possible to share the data before filtering?

Can subjective manual filtering bias the final result?
Decide yourself 👇!

To add-
I agree that to achieve certain precision a min of photons are needed (Pic 1) and filtering helps here
But for me, the resolution is the principle ability to determine a structure
And this is different from what @jelmer_cnossen talks about info/photon and CRLB (Pic 2)

3 dots, continuous triangle, ❤️ or Y?

Suppose u didn't know in advance that the DNA origami below has 3 dots in a triangular array, what would u do?

Sample more to find out the structure or filter bcos we know everything is made of 3 dots in a triangle and improve precision?

🍿 (while we are on 'raw data')

So, I asked for data and got the following reply👇

#MINFLUX people, can u pls stop telling that I don't understand SMLM, can't question a Nobel laureate, etc and answer my questions ❤️?

In btw, I thought #mansplaining happened to females only😜

My response-

To the 1st point, I 'agree' that #MINFLUX is the most photon efficient method to localize molecules although a comparison with other methods is yet to be done (unless I missed a recent publication).

Can we now pls move to the 8-10 points I make in this thread?

To the 2nd point, we don't filter locs to achieve a certain precision.

Of course, if I take events >X photons, I can have an "assured" minimum precision.

But what's the goal here? Filter locs to improve precision and play the resolution game or do biology?

What's your take?

Got to hear a lot of things (via emails, DMs, etc) the last couple of months and these 2 tweets were very timely!

All I can say is THANKS to everyone who has participated in the discussion*, we need more of such open conversations!

(*winters are hard during #COVID19 times 😜!)

A couple of slides from @beniroquai #ImagingONEWORLD talk and a tweet by @TanentzapfLab

Wondering how many of these apply to #STED #MINFLUX?

cc @rita_strack @naturemethods @ScienceMagazine @dfg_public

I think everyone understands the importance of conceptual advancement vs current technical challenges

#MINFLUX is still underdeveloped and needs validation from other independent groups!

It is imp to recall the 4Pi story, probably the most photon efficient method of its time!

Almost the same story for STED!
(still waiting to see some novel biological apps)

https://twitter.com/kirtiprakash25/status/1339486806551777281

The game microscope companies play - launch a new microscope every 5-10 yrs, claim a revolutionary advancement and keep the market moving, but who suffers? Small labs?

In general, scientific papers are hard to write but this one was the hardest I have written!

Happy to see ~300 downloads, ~500 views in one day across different platforms!

A lot of effort went in, please do read and comment!

preprints.org/manuscript/202…

https://twitter.com/kirtiprakash25/status/1358679139595800578

Might come as a surprise but along with Ernst Stelzer and Eric Betzig, Stefan Hell is one of my fav writers!

Some memorable quotes from the MINFLUX papers (Science 2017, Nat Meth 2020)

Another #MINFLUX story:

*Mar 2020 - First email to @naturemethods for access to raw data/codes

*Aug 2020 - Email to @Stefan_W_Hell requesting data/codes with some other concerns

*Since Jan 2021, Nat Meth tells me to expect data any day!

cc @rita_strack @AllisonDoerr14

Curious to see how many #openaccess enthusiasts/PIs/Profs will support the above tweet esp those from EU/UK/US who criticise/mock scientists from eastern Europe/Asia for not providing codes!

This is real punching down, not what is a constructive discussion btw two scientists!

The 1st paper to describe the main physical background of #STED microscopy is by Okhonin in 1987

researchgate.net/publication/27…

Patent:
researchgate.net/publication/27…

researchgate.net/publication/27…

Can Russian scientists/microscopists help with the original patent citation?

cc @HeintzmannLab

And the #STED patent by Stephen C. Baer which the community overlooks!

patents.google.com/patent/US58669…