<div dir="auto"><div><br><span style="font-family:sans-serif;font-size:12.8px">Dear Ariel,</span><div dir="auto" style="font-family:sans-serif;font-size:12.8px"><br></div><div dir="auto" style="font-family:sans-serif;font-size:12.8px">Three check points i can think of.</div><div dir="auto" style="font-family:sans-serif;font-size:12.8px"><br></div><div dir="auto" style="font-family:sans-serif;font-size:12.8px">1: lower your airlock pump time to time 60 seconds. We did that on our Talos and it takes the samples without vacuum break</div><div dir="auto" style="font-family:sans-serif;font-size:12.8px">2. Not sure what your loading tilt angle is. If it is -60or so, feel free to go-to -70. This will save some ln2 in the can</div><div dir="auto" style="font-family:sans-serif;font-size:12.8px">3. Check the o-oring on the holder. Apply some vacuum grease or replace with a new one.</div><div dir="auto" style="font-family:sans-serif;font-size:12.8px"><br></div><div dir="auto" style="font-family:sans-serif;font-size:12.8px">Good luck</div><div dir="auto" style="font-family:sans-serif;font-size:12.8px">Krishna</div><br><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon 24 Feb, 2020, 6:43 PM , <<a href="mailto:3dem-request@ncmir.ucsd.edu">3dem-request@ncmir.ucsd.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Send 3dem mailing list submissions to<br>
<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a><br>
<br>
To subscribe or unsubscribe via the World Wide Web, visit<br>
<a href="https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem</a><br>
or, via email, send a message with subject or body 'help' to<br>
<a href="mailto:3dem-request@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem-request@ncmir.ucsd.edu</a><br>
<br>
You can reach the person managing the list at<br>
<a href="mailto:3dem-owner@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem-owner@ncmir.ucsd.edu</a><br>
<br>
When replying, please edit your Subject line so it is more specific<br>
than "Re: Contents of 3dem digest..."<br>
<br>
<br>
Today's Topics:<br>
<br>
1. Re: Which resolution? (Sjors Scheres)<br>
2. Insertion of Holder 626 (TALAVERA PEREZ Ariel)<br>
<br>
<br>
----------------------------------------------------------------------<br>
<br>
Message: 1<br>
Date: Mon, 24 Feb 2020 09:48:06 +0000<br>
From: Sjors Scheres <<a href="mailto:scheres@mrc-lmb.cam.ac.uk" target="_blank" rel="noreferrer">scheres@mrc-lmb.cam.ac.uk</a>><br>
To: "<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>" <<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>>,<br>
"<a href="mailto:CCPEM@JISCMAIL.AC.UK" target="_blank" rel="noreferrer">CCPEM@JISCMAIL.AC.UK</a>" <<a href="mailto:ccpem@jiscmail.ac.uk" target="_blank" rel="noreferrer">ccpem@jiscmail.ac.uk</a>><br>
Subject: Re: [3dem] Which resolution?<br>
Message-ID: <<a href="mailto:9cb7ea30-8532-66c3-2539-263d3f3e7d72@mrc-lmb.cam.ac.uk" target="_blank" rel="noreferrer">9cb7ea30-8532-66c3-2539-263d3f3e7d72@mrc-lmb.cam.ac.uk</a>><br>
Content-Type: text/plain; charset=UTF-8; format=flowed<br>
<br>
Dear all,<br>
<br>
For the newcomers to the cryo-EM field who may be more familiar with <br>
X-ray crystallography and who may not be familiar with this longstanding <br>
discussion, five observations:<br>
<br>
1) For large numbers of Fourier components per shell, the FSC=0.143 <br>
criterion is correct and equivalent to the FOM=0.5 criterion used in <br>
protein crystallography. From personal experience in RELION it has <br>
worked well in terms of expected behaviour: alpha-helices become tubular <br>
densities around 9-10A, beta-strands become separated at 4.7A, RNA base <br>
pairs at 3.6A, etc.<br>
<br>
2) Marin is correct that there is an argument for a variable threshold <br>
over a fixed one: when the number of (independent) Fourier components <br>
per shell drops (because the shell lies closer to the origin of the <br>
Fourier transform, i.e. is at lower spatial frequency; or in case of <br>
symmetry) one needs to raise the threshold.<br>
<br>
3) However, the amount by how much the threshold changes for typically <br>
cases does, in my opinion, not warrant the language used to make point <br>
2) every other year or so. Please see here <br>
<a href="https://twitter.com/SjorsScheres/status/935182696325763072/" rel="noreferrer noreferrer" target="_blank">https://twitter.com/SjorsScheres/status/935182696325763072/</a> for a plot <br>
of the frequency-dependent behaviour of the 1/2-bit criterion (without <br>
symmetry). It asymptotically approaches 0.172. It was chosen (somewhat <br>
arbitrarily) over the twice as large 1-bit criterion because it was <br>
closer to 0.143. However, for a typical case where the diameter of the <br>
particle D is half the box size L (see Marin's paper here: <br>
<a href="https://www.sciencedirect.com/science/article/pii/S1047847705001292" rel="noreferrer noreferrer" target="_blank">https://www.sciencedirect.com/science/article/pii/S1047847705001292</a>) <br>
<<a href="https://www.sciencedirect.com/science/article/pii/S1047847705001292" rel="noreferrer noreferrer" target="_blank">https://www.sciencedirect.com/science/article/pii/S1047847705001292</a>>the <br>
1/2-bit criterion deviates less from 0.172 than 0.172 itself deviates <br>
from 0.143 for any Fourier shell that is more than 25 shells away from <br>
the origin. So, for typical single-particle reconstructions with box <br>
sizes of say 200-400 pixels used nowadays, and resolutions around say <br>
half-Nyquist, the frequency-dependent threshold will affect the <br>
resolution estimate very little, or at least much less than the <br>
arbitrariness in the 1/2-bit criterion itself. Perhaps someone should <br>
propose to multiply the 1/2-bit curve by 0.83 so that one gets a <br>
frequency-dependent threshold that asymptotically reaches 0.143? ;-)<br>
<br>
4) There are cases where the frequency-dependent threshold does become <br>
more relevant, e.g. for very low resolutions or when using much smaller <br>
box sizes. The latter occurs for example in sub-tomogram averaging or in <br>
(I suspect) methods some people use for local-resolution calculation.<br>
<br>
5) The estimated overall resolution based on whatever FSC-criterion you <br>
favour is just a number. Besides different criteria, there are also <br>
subtle differences in the way different programs calculate the half-maps <br>
and correct for masking effects on the FSC curves. Therefore, don't <br>
obsess over how this number changes in its decimal: what really matters <br>
is the quality (interpretability) of your map.<br>
<br>
Hope that helps,<br>
Sjors<br>
<br>
PS: I am not on CCP4BB, but I would be OK with someone who is forwarding <br>
this message there.<br>
<br>
<br>
<br>
Marin van Heel wrote:<br>
><br>
> Hi Carlos Oscar and Jose-Maria,<br>
><br>
> I choose to answer you guys first, because it will take little of my <br>
> time to counter your criticism and because I have long since been less <br>
> than amused by your published, ill-conceived criticism:<br>
><br>
> ?*/Marin, I always suffer with your reference to sloppy statistics. If <br>
> we take your paper of 2005 where the 1/2 bit criterion was proposed, <br>
> Eqs. 4 to 15 have completely ignored the fact that you are dealing <br>
> with Fourier components, that are complex numbers, and consequently <br>
> you have to deal with random variables that have TWO components, which <br>
> moreover the real and imaginary part are not independent and, in their <br>
> turn, they are not independent of the nearby Fourier coefficients so <br>
> that for computing radial averages you would need to account for the <br>
> correlation among coefficients/*?//<br>
><br>
> I had seen this argumentation against our (2005) paper in your <br>
> manuscript/paper years back. I was so stunned by the level of <br>
> misunderstanding expressed in your manuscript that I chose not to <br>
> spend any time reacting to those statements. Now that you choose to so <br>
> openly display your thoughts on the matter, I have no other choice <br>
> than to spell out your errors in public.<br>
><br>
> All complex arrays in our 2005 paper are Hermitian (since they are the <br>
> FTs of real data), and so are all their inner products. In all the <br>
> integrals over rings one always averages a complex Fourier-space voxel <br>
> with its Hermitian conjugate yielding */ONE/* real value (times two)! <br>
> Without that Hermitian property, FRCs and FSCs, which are real <br>
> normalised correlation functions would not even have been possible. I <br>
> was - and still am - stunned by this level of misunderstanding!<br>
><br>
> This is a blatant blunder that you are propagating over years, a <br>
> blunder that does not do any good to your reputation, yet also a <br>
> blunder that has probably damaged to our research income. The fact <br>
> that you can divulgate such rubbish and leave it out there for years <br>
> for referees to read (who are possibly not as well educated in physics <br>
> and mathematics) will do ? and may already have done ? damage to our <br>
> research.An apology is appropriate but an apology is not enough.<br>
><br>
> Maybe you should ask your granting agencies how to transfer 25% of <br>
> your grant income to our research, in compensation of damages created <br>
> by your blunder!<br>
><br>
> Success with your request!<br>
><br>
> Marin<br>
><br>
> PS. You have also missed that our 2005 paper explicitly includes the <br>
> influence of the size of the object within the sampling box (your: <br>
> ?*/they are not independent of the nearby Fourier coefficients/*?). I <br>
> remain flabbergasted.<br>
><br>
><br>
> On Fri, Feb 21, 2020 at 3:15 PM Carlos Oscar Sorzano <<a href="mailto:coss@cnb.csic.es" target="_blank" rel="noreferrer">coss@cnb.csic.es</a> <br>
> <mailto:<a href="mailto:coss@cnb.csic.es" target="_blank" rel="noreferrer">coss@cnb.csic.es</a>>> wrote:<br>
><br>
> Dear all,<br>
><br>
> I always try to refrain myself from getting into these<br>
> discussions, but I cannot resist more the temptation. Here are<br>
> some more ideas that I hope bring more light than confusion:<br>
><br>
> - There must be some functional relationship between the FSC and<br>
> the SNR, but the exact analytical form of this relationship is<br>
> unknown (I suspect that it must be at least monotonic, the worse<br>
> the SNR, the worse FSC; but even this is difficult to prove). The<br>
> relationship we normally use FSC=SNR/(1+SNR) was derived in a<br>
> context that does not apply to CryoEM (1D stationary signals in<br>
> real space; our molecules are not stationary), and consequently<br>
> any reasoning of any threshold based on this relationship is<br>
> incorrect (see our review).<br>
><br>
> - Still, as long as we all use the same threshold, the reported<br>
> resolutions are comparable to each other. In that regard, I am<br>
> happy that we have set 0.143 (although any other number would have<br>
> served the purpose) as the standard.<br>
><br>
> - I totally agree with Steve that the full FSC is much more<br>
> informative than its crossing with the threshold. Specially,<br>
> because we should be much more worried about its behavior when it<br>
> has high values than when it has low values. Before crossing the<br>
> threshold it should be as high as possible, and that is the "true<br>
> measure" of goodness of the map. When it crosses the threshold of<br>
> 0.143, it has too low SNR, and by definition, that is a very<br>
> unstable part of the FSC, resulting in relatively unstable reports<br>
> of resolution. We made some tests about the variability of the FSC<br>
> (refining random splits of the dataset), trying to put the error<br>
> bars that Steve was asking for, and it turned out to be pretty<br>
> reproducible (rather low variance except in the region when it<br>
> crosses the threshold) as long as the dataset was large enough<br>
> (which is the current state).<br>
><br>
> - @Marin, I always suffer with your reference to sloppy<br>
> statistics. If we take your paper of 2005 where the 1/2 bit<br>
> criterion was proposed<br>
> (<a href="https://www.sciencedirect.com/science/article/pii/S1047847705001292" rel="noreferrer noreferrer" target="_blank">https://www.sciencedirect.com/science/article/pii/S1047847705001292</a>),<br>
> Eqs. 4 to 15 have completely ignored the fact that you are dealing<br>
> with Fourier components, that are complex numbers, and<br>
> consequently you have to deal with random variables that have two<br>
> components, which moreover the real and imaginary part are not<br>
> independent and, in their turn, they are not independent of the<br>
> nearby Fourier coefficients so that for computing radial averages<br>
> you would need to account for the correlation among coefficients<br>
> (<a href="https://www.aimspress.com/fileOther/PDF/biophysics/20150102.pdf" rel="noreferrer noreferrer" target="_blank">https://www.aimspress.com/fileOther/PDF/biophysics/20150102.pdf</a>).<br>
> For properly dealing the statistics, at least one needs to carry<br>
> out a two-dimensional reasoning, including the complex conjugate<br>
> multiplication which is all missing in your derivation, rather<br>
> than treating everything as one-dimensional, real valued random<br>
> variables. Additionally, embedded in your whole reasoning is the<br>
> idea that the expected value of a ratio is the ratio of the<br>
> expected values, that is a 0-th order Taylor approximation of the<br>
> mean of the distribution of a ratio between two random variables.<br>
> Finally, I always find an extreme difficulty to understand the 1<br>
> bit or 1/2 bit criteria, that is, what is the relationship between<br>
> the channel's capacity formula of Shannon<br>
> (<a href="https://en.wikipedia.org/wiki/Shannon%E2%80%93Hartley_theorem" rel="noreferrer noreferrer" target="_blank">https://en.wikipedia.org/wiki/Shannon%E2%80%93Hartley_theorem</a>)<br>
> and our FSC (we do not have any channel through which we are<br>
> "transmitting" our volume, although it is true we have a model<br>
> y=x+n that is the same as in signal transmission, it is not true<br>
> that the average information of a signal is log2(1+SNR); for me,<br>
> the only relationship is that the SNR appears in both formulas,<br>
> FSC and channel capacity, but that does not automatically make<br>
> them comparable and interchangeble). This is not a criticism on<br>
> your work. I think the FSC is a very useful tool to measure some<br>
> properties of the reconstruction process and the quality of the<br>
> dataset (not everything is measured by the FSC) and it also has<br>
> its drawbacks (for instance, systematic errors are rewarded by the<br>
> FSC as they are reproducible in both halves). Moreover, I think<br>
> you are an extremely intelligent person, who I consider a good<br>
> friend, with a very good intuition about image processing and who<br>
> has brought very interesting ideas and methodologies into the<br>
> field. Only that we cannot become crazy about the FSC threshold<br>
> and the reported resolution, as the most interesting part of the<br>
> FSC is not when it is low, but when it is high.<br>
><br>
> I hope I can keep refraining myself in the future :-)<br>
><br>
> Cheers, Carlos Oscar<br>
><br>
> On 2/21/20 6:19 PM, Ludtke, Steven J. wrote:<br>
><br>
>> I've been steadfastly refusing to get myself dragged in this<br>
>> time, but with this very sensible statement (which I am largely<br>
>> in agreement with), I thought I'd throw in one thought, just to<br>
>> stir the pot a little more.<br>
>><br>
>> This is not a new idea, but I think it is the most sensible<br>
>> strategy I've heard proposed, and addresses Marin's concerns in a<br>
>> more conventional way. What we are talking about here is the<br>
>> statistical noise present in the FSC curves themselves. Viewed<br>
>> from the framework of traditional error analysis and propagation<br>
>> of uncertainties, which pretty much every scientist should be<br>
>> familiar with since high-school, (and thus would not be confusing<br>
>> to the non statisticians) ?the 'correct' solution to this issue<br>
>> is not to adjust the threshold, but to present FSC curves with<br>
>> error bars.<br>
>><br>
>> One can then use a fixed threshold at a level based on<br>
>> expectation values, and simply produce a resolution value which<br>
>> also has an associated uncertainty. This is much better than<br>
>> using a variable threshold and still producing a single number<br>
>> with no uncertainty estimate! Not only does this approach account<br>
>> for the statistical noise in the FSC curve, but it also should<br>
>> stop people from reporting resolutions as 2.3397 ?, as it would<br>
>> be silly to say 2.3397 +- 0.2.<br>
>><br>
>> The cross terms are not ignored, but are used in the production<br>
>> of the error bars. This is a very simple approach, which is<br>
>> certainly closer to being correct than the fixed threshold<br>
>> without error-bars approach, and it solves many of the issues we<br>
>> have with resolution reporting people do.? Of course we still<br>
>> have people who will insist that 3.2+-0.2 is better than<br>
>> 3.3+-0.2, but there isn't much you can do about them... (other<br>
>> than beat them over the head with a statistics textbook).<br>
>><br>
>> The caveat, of course, is that like all propagation of<br>
>> uncertainty that it is a linear approximation, and the<br>
>> correlation axis isn't linear, so the typical Normal<br>
>> distributions with linear propagation used to justify propagation<br>
>> of uncertainty aren't _strictly_ true. However, the approximation<br>
>> is fine as long as the error bars are reasonably small compared<br>
>> to the -1 to 1 range of the correlation axis. Each individual<br>
>> error bar is computed around its expectation value, so the<br>
>> overall nonlinearity of the correlation isn't a concern.<br>
>><br>
>><br>
>><br>
>> --------------------------------------------------------------------------------------<br>
>> Steven Ludtke, Ph.D.?<<a href="mailto:sludtke@bcm.edu" target="_blank" rel="noreferrer">sludtke@bcm.edu</a> <mailto:<a href="mailto:sludtke@bcm.edu" target="_blank" rel="noreferrer">sludtke@bcm.edu</a>>> ?<br>
>> ? ? ? ? ? ? ? ? ??Baylor College of?Medicine<br>
>> Charles C. Bell Jr.,?Professor of Structural?Biology<br>
>> Dept. of Biochemistry?and Molecular Biology? ? ? ? ? ? ? ? ? ?<br>
>> ??(<a href="http://www.bcm.edu/biochem" rel="noreferrer noreferrer" target="_blank">www.bcm.edu/biochem</a> <<a href="http://www.bcm.edu/biochem" rel="noreferrer noreferrer" target="_blank">http://www.bcm.edu/biochem</a>>)<br>
>> Academic Director,?CryoEM Core ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?<br>
>> ??(<a href="http://cryoem.bcm.edu" rel="noreferrer noreferrer" target="_blank">cryoem.bcm.edu</a> <<a href="http://cryoem.bcm.edu" rel="noreferrer noreferrer" target="_blank">http://cryoem.bcm.edu</a>>)<br>
>> Co-Director CIBR Center ? ? ? ? ? ? ? ? ?<br>
>> ??(<a href="http://www.bcm.edu/research/cibr" rel="noreferrer noreferrer" target="_blank">www.bcm.edu/research/cibr</a> <<a href="http://www.bcm.edu/research/cibr" rel="noreferrer noreferrer" target="_blank">http://www.bcm.edu/research/cibr</a>>)<br>
>><br>
>><br>
>><br>
>>> On Feb 21, 2020, at 10:34 AM, Alexis Rohou <<a href="mailto:a.rohou@gmail.com" target="_blank" rel="noreferrer">a.rohou@gmail.com</a><br>
>>> <mailto:<a href="mailto:a.rohou@gmail.com" target="_blank" rel="noreferrer">a.rohou@gmail.com</a>>> wrote:<br>
>>><br>
>>> ****CAUTION:*** This email is not from a BCM Source. Only click<br>
>>> links or open attachments you know are safe.*<br>
>>> ------------------------------------------------------------------------<br>
>>> Hi all,<br>
>>><br>
>>> For those bewildered by Marin's insistence that everyone's been<br>
>>> messing up their stats since the bronze age, I'd like to offer<br>
>>> what my understanding of the situation. More details in this<br>
>>> thread from a few years ago on the exact same topic:<br>
>>> <a href="https://mail.ncmir.ucsd.edu/pipermail/3dem/2015-August/003939.html" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/pipermail/3dem/2015-August/003939.html</a><br>
>>> <<a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_pipermail_3dem_2015-2DAugust_003939.html&d=DwMFaQ&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=CZ3YcAV1LVKXsLT0KjCIRby6j3XPA6GqZcOVP3nMyK0&e=" rel="noreferrer noreferrer" target="_blank">https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_pipermail_3dem_2015-2DAugust_003939.html&d=DwMFaQ&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=CZ3YcAV1LVKXsLT0KjCIRby6j3XPA6GqZcOVP3nMyK0&e=</a>><br>
>>> <a href="https://mail.ncmir.ucsd.edu/pipermail/3dem/2015-August/003944.html" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/pipermail/3dem/2015-August/003944.html</a><br>
>>> <<a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_pipermail_3dem_2015-2DAugust_003944.html&d=DwMFaQ&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=oG6lGnei74jC5VVGsfFAdiTpIxrZhs_IH2mH0re5QRM&e=" rel="noreferrer noreferrer" target="_blank">https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_pipermail_3dem_2015-2DAugust_003944.html&d=DwMFaQ&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=oG6lGnei74jC5VVGsfFAdiTpIxrZhs_IH2mH0re5QRM&e=</a>><br>
>>><br>
>>> Notwithstanding notational problems (e.g. strict equations as<br>
>>> opposed to approximation symbols, or omission of symbols to<br>
>>> denote estimation), I believe Frank & Al-Ali and "descendent"<br>
>>> papers (e.g. appendix of Rosenthal & Henderson 2003) are fine.<br>
>>> The cross terms that Marin is agitated about indeed do in fact<br>
>>> have an expectation value of 0.0 (in the ensemble; if the<br>
>>> experiment were performed an infinite number of times with<br>
>>> different realizations of noise). I don't believe Pawel or Jose<br>
>>> Maria or any of the other authors really believe that the<br>
>>> cross-terms are orthogonal.<br>
>>><br>
>>> When N (the number of independent Fouier voxels in a shell) is<br>
>>> large enough, mean(Signal x Noise) ~ 0.0 is only an<br>
>>> approximation, but a pretty good one, even for a single FSC<br>
>>> experiment. This is why, in my book, derivations that depend on<br>
>>> Frank & Al-Ali are OK, under the strict assumption that N is<br>
>>> large. Numerically, this becomes apparent when Marin's half-bit<br>
>>> criterion is plotted - asymptotically it has the same behavior<br>
>>> as a constant threshold.<br>
>>><br>
>>> So, is Marin wrong to worry about this? No, I don't think so.<br>
>>> There are indeed cases where the assumption of large N is<br>
>>> broken. And under those circumstances, any fixed threshold<br>
>>> (0.143, 0.5, whatever) is dangerous. This is illustrated in<br>
>>> figures of van Heel & Schatz (2005). Small boxes, high-symmetry,<br>
>>> small objects in large boxes, and a number of other conditions<br>
>>> can make fixed thresholds dangerous.<br>
>>><br>
>>> It would indeed be better to use a non-fixed threshold. So why<br>
>>> am I not using the 1/2-bit criterion in my own work? While<br>
>>> numerically it behaves well at most resolution ranges, I was not<br>
>>> convinced by Marin's derivation in 2005. Philosophically though,<br>
>>> I think he's right - we should aim for FSC thresholds that are<br>
>>> more robust to the kinds of edge cases mentioned above. It would<br>
>>> be the right thing to do.<br>
>>><br>
>>> Hope this helps,<br>
>>> Alexis<br>
>>><br>
>>><br>
>>><br>
>>> On Sun, Feb 16, 2020 at 9:00 AM Penczek, Pawel A<br>
>>> <<a href="mailto:Pawel.A.Penczek@uth.tmc.edu" target="_blank" rel="noreferrer">Pawel.A.Penczek@uth.tmc.edu</a><br>
>>> <mailto:<a href="mailto:Pawel.A.Penczek@uth.tmc.edu" target="_blank" rel="noreferrer">Pawel.A.Penczek@uth.tmc.edu</a>>> wrote:<br>
>>><br>
>>> Marin,<br>
>>><br>
>>> The statistics in 2010 review is fine. You may disagree with<br>
>>> assumptions, but I can assure you the ?statistics? (as you<br>
>>> call it) is fine. Careful reading of the paper would reveal<br>
>>> to you this much.<br>
>>><br>
>>> Regards,<br>
>>> Pawel<br>
>>><br>
>>>> On Feb 16, 2020, at 10:38 AM, Marin van Heel<br>
>>>> <<a href="mailto:marin.vanheel@googlemail.com" target="_blank" rel="noreferrer">marin.vanheel@googlemail.com</a><br>
>>>> <mailto:<a href="mailto:marin.vanheel@googlemail.com" target="_blank" rel="noreferrer">marin.vanheel@googlemail.com</a>>> wrote:<br>
>>>><br>
>>>> ?<br>
>>>><br>
>>>> ***** EXTERNAL EMAIL *****<br>
>>>><br>
>>>> Dear Pawel and All others ....<br>
>>>><br>
>>>> This 2010 review is - unfortunately - largely based on the<br>
>>>> flawed statistics I mentioned before, namely on the a<br>
>>>> priori assumption that the inner product of a signal vector<br>
>>>> and a noise vector are ZERO (an orthogonality assumption).?<br>
>>>> The (Frank & Al-Ali 1975) paper we have refuted on a number<br>
>>>> of occasions (for example in 2005, and most recently in our<br>
>>>> BioRxiv paper) but you still take that as the correct<br>
>>>> relation between SNR and FRC (and you never cite the<br>
>>>> criticism...).<br>
>>>><br>
>>>> Sorry<br>
>>>> Marin<br>
>>>><br>
>>>> On Thu, Feb 13, 2020 at 10:42 AM Penczek, Pawel A<br>
>>>> <<a href="mailto:Pawel.A.Penczek@uth.tmc.edu" target="_blank" rel="noreferrer">Pawel.A.Penczek@uth.tmc.edu</a><br>
>>>> <mailto:<a href="mailto:Pawel.A.Penczek@uth.tmc.edu" target="_blank" rel="noreferrer">Pawel.A.Penczek@uth.tmc.edu</a>>> wrote:<br>
>>>><br>
>>>> ?Dear Teige,<br>
>>>><br>
>>>> I am wondering whether you are familiar with<br>
>>>><br>
>>>><br>
>>>><br>
>>>> Resolution measures in molecular electron microscopy.<br>
>>>><br>
>>>> Penczek PA. Methods Enzymol. 2010.<br>
>>>><br>
>>>><br>
>>>> Citation<br>
>>>><br>
>>>> Methods Enzymol. 2010;482:73-100. doi:<br>
>>>> 10.1016/S0076-6879(10)82003-8.<br>
>>>><br>
>>>><br>
>>>> You will find there answers to all questions you asked<br>
>>>> and much more.<br>
>>>><br>
>>>> Regards,<br>
>>>> Pawel Penczek<br>
>>>><br>
>>>><br>
>>>> Regards,<br>
>>>> Pawel<br>
>>>> _______________________________________________<br>
>>>> 3dem mailing list<br>
>>>> <a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a> <mailto:<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>><br>
>>>> <a href="https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem</a><br>
>>>> <<a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_mailman_listinfo_3dem&d=DwMFaQ&c=bKRySV-ouEg_AT-w2QWsTdd9X__KYh9Eq2fdmQDVZgw&r=yEYHb4SF2vvMq3W-iluu41LlHcFadz4Ekzr3_bT4-qI&m=3-TZcohYbZGHCQ7azF9_fgEJmssbBksaI7ESb0VIk1Y&s=XHMq9Q6Zwa69NL8kzFbmaLmZA9M33U01tBE6iAtQ140&e=" rel="noreferrer noreferrer" target="_blank">https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_mailman_listinfo_3dem&d=DwMFaQ&c=bKRySV-ouEg_AT-w2QWsTdd9X__KYh9Eq2fdmQDVZgw&r=yEYHb4SF2vvMq3W-iluu41LlHcFadz4Ekzr3_bT4-qI&m=3-TZcohYbZGHCQ7azF9_fgEJmssbBksaI7ESb0VIk1Y&s=XHMq9Q6Zwa69NL8kzFbmaLmZA9M33U01tBE6iAtQ140&e=</a>><br>
>>>><br>
>>> _______________________________________________<br>
>>> 3dem mailing list<br>
>>> <a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a> <mailto:<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>><br>
>>> <a href="https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem</a><br>
>>> <<a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_mailman_listinfo_3dem&d=DwMFaQ&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=TeEhUNYC5v59HGWMrPQCMaGK5opuX-NIG2mJvGLuiKA&e=" rel="noreferrer noreferrer" target="_blank">https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_mailman_listinfo_3dem&d=DwMFaQ&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=TeEhUNYC5v59HGWMrPQCMaGK5opuX-NIG2mJvGLuiKA&e=</a>><br>
>>><br>
>>> _______________________________________________<br>
>>> 3dem mailing list<br>
>>> <a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a> <mailto:<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>><br>
>>> <a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_mailman_listinfo_3dem&d=DwICAg&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=TeEhUNYC5v59HGWMrPQCMaGK5opuX-NIG2mJvGLuiKA&e=" rel="noreferrer noreferrer" target="_blank">https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.ncmir.ucsd.edu_mailman_listinfo_3dem&d=DwICAg&c=ZQs-KZ8oxEw0p81sqgiaRA&r=Dk5VoQQ-wINYVssLMZihyC5Dj_sWYKxCyKz9E4Lp3gc&m=UWn2RUCMENrXjn3JLSwlIU6Zmp_JYnRrXesjtsM1u2E&s=TeEhUNYC5v59HGWMrPQCMaGK5opuX-NIG2mJvGLuiKA&e=</a><br>
>><br>
>><br>
>> _______________________________________________<br>
>> 3dem mailing list<br>
>> <a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a> <mailto:<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>><br>
>> <a href="https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem</a><br>
><br>
><br>
><br>
> _______________________________________________<br>
> 3dem mailing list<br>
> <a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a><br>
> <a href="https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem</a><br>
<br>
-- <br>
Sjors Scheres<br>
MRC Laboratory of Molecular Biology<br>
Francis Crick Avenue, Cambridge Biomedical Campus<br>
Cambridge CB2 0QH, U.K.<br>
tel: +44 (0)1223 267061<br>
<a href="http://www2.mrc-lmb.cam.ac.uk/groups/scheres" rel="noreferrer noreferrer" target="_blank">http://www2.mrc-lmb.cam.ac.uk/groups/scheres</a><br>
<br>
<br>
<br>
------------------------------<br>
<br>
Message: 2<br>
Date: Mon, 24 Feb 2020 13:13:04 +0000<br>
From: "TALAVERA PEREZ Ariel" <<a href="mailto:Ariel.Talavera@ulb.be" target="_blank" rel="noreferrer">Ariel.Talavera@ulb.be</a>><br>
To: "<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>" <<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a>><br>
Subject: [3dem] Insertion of Holder 626<br>
Message-ID: <<a href="mailto:b31b7e8d-9bca-4722-1616-825fedfbbd68@ulb.be" target="_blank" rel="noreferrer">b31b7e8d-9bca-4722-1616-825fedfbbd68@ulb.be</a>><br>
Content-Type: text/plain; charset="utf-8"<br>
<br>
Dear all,<br>
<br>
I am using a Gatan 626 holder on a side entry TALOS microscope. I am <br>
getting problems with very frequent crystalline ice formation on the <br>
grids. Some times I get a mix of vitreous and crystalline water, but <br>
must of the time I only get crystalline water. I have been extra careful <br>
during sample preparation, and transferring the grind on to the holder, <br>
and I have done 48 hours holder bake out before insertion, but I still <br>
get ice on the grid.<br>
<br>
I monitored the temperature of the holder during the insertion process, <br>
including 3 minutes for the Airlock pump. During those 3 minutes the <br>
temperature rises up to -152.2. Can this be the reason of the unwanted <br>
ice formation? How high can be the temperature of the holder to ensure a <br>
nice vitreous water?<br>
<br>
I am also getting vacuum problems while inserting the holder. This <br>
process has resulted to be extremely, extremely sensitive. After the 3 <br>
minutes Airlock pumping (I cannot give longer time because of the <br>
aforementioned temperature issue) as soon as I open the column port the <br>
vacuum either rises to 99 Log or completely crashes the column vacuum. <br>
This has happened with two different holders either cooled with liquid <br>
nitrogen or at room temperature. So, my question: Is this insertion <br>
process always that? picky or it can be that there is something wrong <br>
with the holders/Airlock pump?<br>
<br>
Thanks a lot in advanced.<br>
<br>
Best regards.<br>
<br>
Ariel<br>
<br>
-- <br>
Ariel TALAVERA PEREZ, PhD<br>
Center for Microscopy and Molecular Imaging (CMMI)<br>
Laboratory of Microscopy<br>
Universit? Libre de Bruxelles<br>
Rue Adrienne Bolland, 8<br>
B-6041 Gosselies, Belgium<br>
Phone +32 (0)26509866<br>
<br>
<br>
------------------------------<br>
<br>
Subject: Digest Footer<br>
<br>
_______________________________________________<br>
3dem mailing list<br>
<a href="mailto:3dem@ncmir.ucsd.edu" target="_blank" rel="noreferrer">3dem@ncmir.ucsd.edu</a><br>
<a href="https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem" rel="noreferrer noreferrer" target="_blank">https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem</a><br>
<br>
<br>
------------------------------<br>
<br>
End of 3dem Digest, Vol 150, Issue 56<br>
*************************************<br>
</blockquote></div></div></div>