BioImaging Centre. Dan Clare : Single Particle Data collection

Transcription

1 ebic: electron BioImaging Centre Dan Clare : Single Particle Data collection Wellcome Trust/MRC/BBSRC Strategic Award Helen Saibil, Kay Grünewald, David Stuart, Gerhard Materlik

2 3 FEI transmission electron microscopes Microscopes at Diamond Krios 1 and Krios 2 (300 KeV) are both offered as part of the user program Talos (200 KeV) installation is almost complete Both Krios microscopes are equipped with the latest generation of direct electron detectors - Quantum-K2 (400 fps 40 fps) and Falcon II (17 fps) All microscopes have Volta Phase Plates - Not integrated into single particle software yet!

3 Data collection at Diamond As of the 02/09/ external visits, 62 unique, TB of data generated Averaging just over 2TB/ 48 hr session with the Falcon II direct electron detector single particle only The first facility in the world to have the bioquantum-k2 Summit detector integrated with EPU and averaging 2.4 Tb/ 48 hr session Multiple ~3Å reconstructions comprising multiple submitted/accepted manuscripts. Two reconstructions at better than 3Å In the *first year* of operation we will have provided 220 days for external users, over delivering by ~40%... (Krios 1)

4 Typical User Day Session starts at : (most people are early!) : Load grids into the auto grid cartridges and Krios cassette : Load grids into the Krios : Survey grids to determine the best grid : Collect grid atlas image : Setup EPU, microscope alignments, detector gain refs. etc. 3 4: Check images, assess CTF etc. Automatic collections for ~40 hours. Staff monitor remotely (but so can you!).

5 EPU Automated Image Acquisition Atlas Grid Square Hole Summed High resolution image Setup the template for image acquisition: Focus, Drift Measurement and Exposure(s) area. This template will be automatically repeated for every selected hole. Images provided by Sonja Welsch (FEI)

6 Things to Consider Before Your Single Particle Data Collection at Diamond o How many grids you want to load Less is more? o Detector Integrating versus counting? o Magnification Number of particles, resolution? o Hole size Speed of collection? o Number of frames, Dose & Defocus Microscope stability and object size? o Data Data assessment and how do I get it back?

7 How many Grids Maximum of 12 grids can be loaded into the Cassette We recommend loading only 8 grids as this will already take around 1 hour. The more grids that need to be screened the less time you will have for collection! You can bring a backup sample and I would recommend loading 6 grids for your primary sample and 2 for your backup sample. Make sure that all your samples are on the ERA. Otherwise they will not be loaded! We also prefer to load only one cassette per 48hour session!

8 Gatan Website Detector Choice Integrating vs Counting Electron enters detector Signal scattered Charge collects in each pixels Events are reduced to highest charge pixels Events are localised with sub-pixel accuracy Falcon II integrating Optimal dose rate 55e/p/s shorter exposure K2 Counting? K2 Super resolution counting? Optimal dose rate < 5e/p/s longer exposure Cheng et al, 2013 McMullan et al, 2014

9 Detector Choice Integrating vs Counting McMullan et al, 2014 Advantages of Integration (Falcon II) Works at a range of electron doses, therefore short exposure times (1-2 seconds) can be used at most magnifications Potentially better DQE at high spatial frequencies than K2 Advantages of counting/super-resolution counting (K2) Much higher DQE at lower spatial frequencies (removes the random energy deposition of detected electrons and read noise) Improve particle alignment Gatan Website Super-resolution can recover information past the Nyquist frequency of the original image (Removes aliasing)

10 Detector Choice - Recommendation For large viruses/protein complexes (above 1MDa) it may be advisable to use the Falcon II as you probably don t need the increased DQE of the K2 at lower spatial frequencies but do need the increased DQE at higher spatial frequencies The Falcon II can collect more images per hour than the K2 e.g. with our fastest data collection strategies we can get 100 movies per hour on the Falcon II versus 75 on the K2 Potentially 1000 more movies in a 48 hour session For Heterogeneous samples it could be an advantage to use the Falcon II More data. However for smaller complexes (less than 500kDa) it would be better to use the K2 For the increased DQE at lower spatial frequencies. Super-resolution gives better DQE at higher spatial frequencies than normal counting mode but is significantly slower

11 Magnification What resolution do you want to achieve? Both the falcon II and the K2 will provide information at the Nyquist frequency of the image. The most common magnifications used on both the Falcon II and K2 give approximately 1.4 or 1.1 Å/pix Using a lower magnification may not give you more particles as you can not take as many movies per hole (next slide) Lower magnifications on the K2 require longer exposure times e.g 5e/p/s in counting will be equivalent to 4.2 e/å 2 /s at 1.1 Å/pix compared to 2.6 e/å 2 /s at 1.4 Å/pix - translates to an increase in exposure time by 40 % to achieve the same total dose. McMullan et al, 2014

12 Hole Size Lacey 0.6uM holes 1.2uM holes 2uM holes 3.5uM holes Falcon II 75 IM/hr 1 IM/hole = 45 IM/hr 3 IM/hole = 65 IM/hr 5 IM/hole = 75 IM/hr* * K2 55 IM/hr 1 IM/hole = 30 IM/hr 3 IM/hole = 55 IM/hr 5 IM/hole = 65 IM/hr 8 IM/hole = 75 IM/hr *Averages and based on our standard imaging parameters of approximately 2 second exposures with the Falcon II (35 frames) and 8 second exposures with the K2 (20 frames). *Increasing the number of frames or the exposure time can reduce the number of IM/hr. *Super-resolution gives approximately half the number of IM/hr as counting Ultrafoil gold grids can be used but require a carbon grid loaded for microscope alignment

13 Number of frames, Dose & Defocus range The krios is a very stable microscope! So you may not require as many frames as you would think. Routinely we use a 2 second exposure with 17 frames per second on the Falcon II and an 8 second exposure with 20 frames on the K2 (1Gb per movie) How many electrons will you need? We tend to routinely use between electrons, However for larger complexes this may not be necessary and for smaller complexes this may not be enough! With our standard frame and dose rates you get ~ 1.5 e/frame/s on Falcon II and 2e/frame/s on the K2 With the falcon II you can use shorter frame times initially May help with beam induced movement! Defocus range is dependent on complex size e.g For large virus use a defocus range from -0.6 to -2 um (0.3 to 0.5 steps) compared to -1.8 to -3.8 for a 250 kda protein complex You need to be able to see your complex!

14

15 Automated Pipeline (Quality control) We are currently in the process of implementing an automated pipeline that will allow the user to align their movies, estimate defocus values, pick particles (user provided templates), align and 2D classify However at the moment these operations have to be done manually - dedicated user office for this We would recommend that the users at least check that the defocus range they have asked for is correct Getting your data home Bring USB drives as we can copy as we can copy the data when it is acquired Alternatively you can FTP/Globus to send them back to your home institution!

16 Commissioning tests using GroEL Falcon II 1.4 Å/pixel 7 frames in 1.5 seconds (short initial frames and along final frame) Total dose ~ 40 electrons Movie alignment IMOD 33,000 particles in final reconstruction (processed with Relion) FSC = 4.1 Å Quantum-K2 Krios Å/pixel (0.503 SR) 16 frames in 4 seconds using at counting rate of 8e/Å2/s, superresolution mode Total dose ~ 40 electrons Movie alignment IMOD 17,400 particles in the final reconstruction (Processed with Relion after image binning) FSC = 3.3 Å Quantum-K2 Krios Å/pixel 20 frames in 10 seconds using at counting rate of 4e/Å2/s Total dose ~ 40 electrons Movie alignment Motioncorr 38,000 particles in the final reconstruction (Processed with Relion) FSC = 4 Å *Old protein*

17 Falcon II K2 (SR) K2 (C)

18 People ebic staff: Alistair Siebert and Corey Hecksel. Diamond light source staff: Jean Lane, Alun Ashton, Kevin Savage, Nick Rees, Michelle Bennett, Alison Roblin, the EHC s and others. Alistair Siebert, Dan Clare, Alex Buzduga, Sonja Welsch, Alan Boswell FEI installation and application support team: Alan Boswell, Alex Buzduga, Alberto Gonzalez, Sonja Welsch, Felix de Haas, Sacha de Carlo and more. Gatan Installation: Liam Spillane. ebic management: Dave Stuart, Kay Grünewald, Helen Saibil & Martin Walsh.

19 The NEW Direct Electron Detectors (DED) Silicon MAPS detectors 30μm Sensor pixel size 5 or 6 μm - Electrons are directly detected by the pixel electronics of the detector. The DED are also made of silicon which reduces the spread of electrons to neighboring pixels. The substrate underneath is back thinned to stop electrons being scattered back to the epilayer. The charge from the pixels are read individually so there is effectively no dead time for the detector, therefore the detector can run at very high speeds. Movie from Ben Bammes

20 The aliasing effect occurs when there are not enough pixels to capture all of the detail in the picture: changes - details - are occurring in the image for which there are simply not enough available pixels to record faithfully. Normally, we would choose a high enough resolution device to ensure that we have enough pixels to capture the detail in image to the required accuracy. If there should not be a sufficient number, we would simply expect to see less detail. However, this is not the case: sampling theory tells us that the situation is actually much worse and states that if we do not have a sufficient number of pixels, the image is permanently degraded. Gatan Website Nikon Website

21 Both reconstructions were nearly identical and It was possible to place 90 % of the side chains from TMV in density at this resolution! DE20 DED reconstructions RNA Gatan K2 RNA

22 (Lu et al., 2014)

23 Data collection statistics Number of unique groups Allocated time *14 groups from Cambridge, 10 groups from Oxford, 6 groups from Birkbeck, 5 groups from Imperial, 4 groups from Manchester and Bilbao, 3 groups from Leeds, 2 groups from Edinburgh, the Crick and Dundee, 1 Group from Diamond, Warwick, Madrid, Bristol, Leicester, Helsinki, Sheffield, Stockholm, Virginia and SPring8