These instructions allow to perform a two-dimensional analysis and to create a 3D structure from images obtained on the electron microscope. In the random conical method, tilt pairs are selected with the command "particles" in WEB and windowed. The zero-degrees images are subjected to 2D alignment towards a reference. Multivariate classification analysis is applied to discern different groups. At this point, either a 2D analysis -including determination of the 2D resolution and four-fold symmetrizing a 2D average- or a 3D reconstruction can be performed. In the latter case, the 3D is obtained using the rotational parameters and the tilt single particles for a selected group. Procedures to determine the resolution of a volume, to four-fold symmetrize a volume, to adjust size and other procedures are included. Appendix with Projection Matching Methods.

updated 9/28/2000, Montserrat Samso
written 5/20/97, Montserrat Samso







Two-Dimensional Alignement with Reference and Preparation of Tilted Images
1.Micrographs in the scanner format  can be converted to Spider
	format using b01.cpf
	This creates files in SPIDER format, one with original size and
 	another reduced (use an even number, e.g.4). 
	In that copy operation, SPIDER accepts a different file extension
	(which will become output extension) than the input extension (which in Perkin 
	Elmer is always img). Input extension (img) should be in capital
	letters.
	To convert micrographs scanned with the new scanner (HI-SCAN), use  procedure
	b01.cpr 
	
	Check power spectrum and histogram of a 1000x1000 window using 
	b01.scn
	Display with gnuplot. In this program, type: load 'plot1' 
	or load 'plot2'.

2.Choose command PARTICLES in WEB and select pairs from the 
	reduced size micrographs. Do the
	option Fit angles regularly, and check fitted positions.
	Choose 10-15 windows in representative areas where there are no particles, using
	option Background. 

	Alternatively, choose the command MARKERS using unreduced
	micrographs.

3.Window images out from micrographs. 
	For 2D analysis:
	For micrographs scanned with "hi scan" and particles picked with command tilted particles, 
	with ramp and correction respect histogram 
	of noise file: use b03.win
	(runs procedure @window2db).
	For micrographs scanned with "hi scan" and particles picked with command "markers": 
	use b04.win
	(runs procedure @window2dc).
	For micrographs scanned on the Perkin Elmer with inversion of 
	contrast, ramp, and correct respect histogram of noise file:
	use b02.win
	(runs the procedure @window2da).
	For windowing without contrast inversion, ramp or contrast enhancement:
	use b01.win
	(runs the procedure @window2d).

	For windowing several micrographs at the same time (hi scan,
	ramp, histogram correction, no inversion, markers)
	Will give a list with: micrograph #,  3 particles/micrograph,
	first particle, last particle for that micrograph
	b05.win

	Plain windowing without any further modification:
	b06.win
	windowf.win

	For 3D analysis, use b01.twi for tilted images 
	and b01.uwi for untilted images. These will 
	create a continuous image series and correct for the 
	different background in particles from different micrographs. A
	rotation equivalent to the phi angle is performed. The header of
	each image contains the tilt angle.
	
5.Centration of particles by cross-correlation with reference. 
	The reference can be a blob or a circularly symmetrized and 
	low pass filtered version of an existing 2D average of the particle. 

	For integer shifts, external reference use
	b12.ori
	which runs the procedure @center

	For integer shifts, blob, use
	b14.ori
	which runs the procedure @centerb.ori
		
		
	To adjust the size of an average obtained using different magnification,
	read this. Then, do b03.siz
	which runs the procedure @sizeplus	

	or b01.siz
	which runs the procedure @sizeminus	
	
6.Orientational alignment towards a given reference.
	a. Mask and low pass filter a representative average of the dataset interactively using 
	to create the reference using 
	b03.ori which runs
	@rcal1pre. 
	Check the average of the outcoming reference (org999). If its value is less than 0, make it larger than
	zero using AR.	
	b. Run b89.ori
	This procedure file calls the procedure rcals1hp.ori

7. Second round of orientational alignment towards average obtained
	for this dataset in previous step.
	a. Mask and low pass filter a representative average of the dataset interactively using 
	to create the reference using @rcal1pre. 
	Check the average of the outcoming reference (org999). If its value is less than 0, make it larger than
	zero using AR.	
	b. Run b90.ori
	This procedure file calls the procedure 
           rcals1hp.ori

8.Sum alignment of document files. No mirroring.
	b01.sap

9.Apply alignment document file to original files. No mirroring.
	b01.rts

10.For tilted pairs, do this step. For 2D analysis, go to step 11.
	Center tilted images respect their 0 deg
	pairs and get theta angles on the headers.
	
	a. Create average of resulting oriented images using SPIDER operation AS
	b. Mask and low pass obtained average interactively using @rcal1pre.
	c. b06.ori.
	This procedure file calls the following procedures:
	rclabel2mod
	rctcenh

11.Check that the final aligned images are correct (e.g.,
	displaying the averages and variances) and delete all intermediate files.

12.Remove unaligned particles and junk, and copy the good (aligned) ones to
	another directory in a continuous file series.
	a.In WEB use the Categorize command by
	displaying the image files and manually clicking on each bad particle 
	under category 2 and saving the file numbers into a document file. It 
	is better to repeat the same process for the tilted images and saving the
	file numbers in the same document.
	Run b07.sel

	Alternatively, use Categorize command in WEB clicking on
	the good (aligned) particles under category 1. Then, run operation DOC SORT
	followed by b05.sel
	
Go to (classification) step.







Other Procedures




Procedure to know the ratio of projections
	that have moved more than x degrees between an iteration and the next,
	during the projection-matching refinement.
	Corrected for 4-fold symetry. 
	b24.lng
	lang4s





Multivariate Classification Analysis of the 0 Degree Images





13.Low pass filter the 0 deg projections.
	b85.fqu

14.Create a mask.
	BC of  a final average low pass box 10,10, filtration 0.5
	FS and get average value
	TH M above average value
	BC of the resulting mask 
	FS and get average value
	TH M above average value, to obtain a central 
	white mask (ones) in a black (zeroes) background.

15.First part of CORAN. This file has to be run in the 
	directory where we want the IMG file.
	b01.cas
	
16.Second part of CORAN. This file has to be run in the directory 
	where the IMG file is.
	b03.cas
	If CL CLA is not working, use
	b07.cas

17.Check the first 8 factors (specially clockwise/anticlockwise) with
	the procedure evalcoran,
	b01.eva

18.Get different groups of particles.
	a.In WEB, use command Dendrogram and check a good threshold to
	differentiate between clockwise and anticlockwise projections.
	b.Run CL HE to create selection files. 
	c.If it's necessary to join 2 groups but excluding some other 
	and it's not possible doing it by CL HE, join the selection
	files by editing them. Use a procedure file to add a constant to 
	the key number.
	b03.sel
	d.Run AS DC to  get averages for every group. This will
	be useful to decide which averages to join. To test resolution, check 
	the Two-dimensional analysis chapter.
	b03.asd
	e.Run AS R to get averages that can be used in the t-test,
	useful when two-dimensional averages are compared.
	b04.asr

19.Create 2 parallel directories for clockwise/anticlockwise
	particles.
	b05.sel





At this point either  Two-dimensional analysis 
 or  Three Dimensional Reconstruction  can be
performed.


	

Two-dimensional Analysis



20. In order to decide which classes to merge, it can be useful to 
	get the resolution for different ways of merging classes.

	Make selection files for "odd" and "even" numbers for a chosen class
	using b22.ode.
	which runs de procedure the procedure 
	p_oddeven.ode.

21.	Get average for "odd" particles.
	b05.asr

22.	Get average for "even" particles.
	b06.asr

23.	Find resolution
	b07.rf2
	The second column indicates the phase residual. Take the value of the first
	column at which the 2nd column goes above 45.

	The third column indicates the fourier ring correlation. Get the value of the first
	column at which the values on the 3rd column go below 0.5.
	Alternatively, check at which point this curve intersects with
	the curve from the 4th column (Critical Fourier Ring Correlation).

	To calculate the resolution in real space, divide pixel size by the fourier 
	radius (value obtained from the first column at the above cutoff values).	

	Resolution can also be obtained using the following procedure:
	b03.rf2.

24.	Once the optimal group of particles is chosen, get average and variance of
	this group of particles.
	b08.asr
  


25.Four fold symmetrize averages using the procedure
	rav2dal,
	b01.4fo.	

26.Comparison of two averages with the same overall conformation 
	which differ in a certain area, using
	DR DIFF.
	a.Obtain reference using @rcal1pre.
	b.Orient the four-fold symmetrized images.
	b14.ori.
	c.Create a mask as described in point 11.
	d.Normalize images.
	b01.cor.
	e.Perform DR DIFF.
	b01.drd.
	f.Four fold symmetrize.
	b01.4fo.	
	h.Threshold this slice to select positive differences. TH M.
	Use the option "below". A good value to start is the average
	value of the image.

 Signifficance test
	Also threshold signifficance map at 98% confidence level and
	4-fold symmetrize the eptt map
	b04.ept
	


	

3D Reconstruction Using Weighted Back Projection Algorithms


27.Remove unaligned particles and junk, and copy the good (aligned) ones to
	another directory in a continuous file series.
	a.In WEB use the Categorize command by
	displaying the image files and manually clicking on each bad particle 
	under category 2 and saving the file numbers into a document file. It 
	is better to repeat the same process for the tilted images and saving the
	file numbers in the same document.
	b.In SPIDER, run operation AT IT to put the particle
	numbers in a consecutive order. Check the total number of bad
	particles.
	c. Run b06.sel

28.Create 2 parallel directories for clockwise/anticlockwise
	particles.
	Create a selection file for each (clockwise/anticlockwise)
	series.
	b02.sel
	Copy particles to each directory using a continuous series and
	perform 3 rounds of BP XY and 
	altovol
	using procedure file b01.rec.

29.Calculate total shift and compute 3D using back projection
	algorithms.
	b02.rec

30.Take a quick look at the reconstruction (low pass filter, invert
	sign).
	b03.rec.


31.Four fold symmetrize volume. This procedure file calls the procedures
	rav3dal
	cmask3d3
	b04.rec


32.Split select file used in 3D into two separate select files to be
	used in the following two 3D reconstructions for comparative 
	purposes.
	b05.rec

33.Compute two 3D reconstructions of the odd and even particles for clockwise set.
	b06.rec

34.Compute two 3D reconstructions of the odd and even particles for anticlockwise set.
	b07.rec

35.Compare the two half volumes
	b08.rec	

36.In UNIX, use gnuplot to view the resulting curve
	Plot 'rfdoc' using 3:5 with lines







3D Reconstruction Using Projection Matching Methods

Scan images using Hi-Scan 

b01.cpr 
	Convert scanned micrograph file to SPIDER image file.

b22.pws 
	Generate the CTF estimate of the power spectrum.
        for first 8 micrographs 
	p_power.pws 
	
Estimate how good is the micrograph by checking the power spectrum.

promat/64_100.gif
 

Pick particles using command markers
	Only pick good ones, leave out areas of drift
	

window2dc.win 
	Window out particles 

b02.int
	Interpolation to 75x75 of uncentered images

compress full*

categorize and make goodones001.
	
AT IT 

b14.pjq	
	create reference projections

	reference projections: use Iptxa- control
	mkdir /net/ithaca/usr21/samso/RyR1cntiptxa_49_75_refprj
	cp /net/ithaca/usr21/samso/samso/cnt_sa_fh_mref_2/refi/vol004.tox
	Its resolution is 30 A but I'll filter it to 35 A since I expect
	a different conformation and I don't want to condition the
	initial volume too much fq np Gaussian lo-pass (7.7/0.22=35)
	0.22, 0.03 = volref.cam
	cp /net/ithaca/usr21/samso/RyR1_49_75_refprj_ip/selectref.tox 
	cp /net/ithaca/usr21/samso/ang.tox 	

cop.exe 
	if file extension needs to be changed

run a test apmq in spi90  to see how many entries has the current version

b16.amq 	
	Align the particles to the reference projections. 
        This is a multireference alignment of an image series.
	run with spi90 on ithaca (on 6/10/99 it is spider4.285)

b14.ang	
	Create a file containing particle numbers and reference angles for
        each particle to be used to compute the 3D reconstruction.
 	Runs the procedure
        p_angles.ang
	output= angles001

b15.rot	
	Rotate particles according to alignment parameters.
        Any particles that correspond to reference projections 49-96
        are also mirrored since projections 49-96 are mirror images
        of the first 48. 
	Runs the procedure
        p_rotate.rot
	output=proj1/imbip/imb****

Compare aligned particles to reference projections.
        Create a file that reports the number of particles in each
        reference view.  Create a set of files (one for each 
        reference view) that reports associated particle numbers
        and correlation coeffiecients. The correlation coefficient 
        describes the relative similiarity of the particle to the 
        reference projection. Enter 0 for the correlation coefficient
        threshold to include all particles.
        
   	group.f
        
        To begin this Fortran program, select the path to group.exe
        below, and place the selection into your UNIX window. 

        /net/saba/usr1/pawel/useful-fortran-programs/group.exe

	mkdir sel0 for the selection groups at threshold 0

        You will be prompted to enter the following parameters:

        Data code:ext
        CCC threshold: 0
        Number of reference images: 48
        APMQ document file: apmq001
        Enter template for selection doc: sel0/selecgroup001
        Document file: how_many001

	the different selection files are called selecgroup*,
	and the document with the amount of particles which fell on each of the 
	48 projections is called how_many001

	check the different groups with montage from document file, and
	Identify a minimum correlation coefficient that describes true particles 
	as opposed to erroneously selected particles

	display the document file using gnuplot. To do that, type
	gnuplot
	plot 'how_many001.cam' using 3:5 with boxes 
	set yrange [0.0:50] 
	if yrange needs to be changed

	Decide how many top versus side views will enter the 3D
	reconstruction program.

b19.dis	
	Generate an image with 2D distribution of angular directions.
        This program creates a large circular plot in which smaller 
        circles that represent the 48 angular groups are placed. The
        radius of these 48 circles are proportional to the number of 
        particles in each group.        
        output: cndis001
	
	Display this file in WEB

	Check that the side views are represented enough to proceed with
	the 3D reconstruction method.

b19.eqp	
	To sort document files according to cross-corrrelation and 
	cut down amount of particles to a maximum number in order 
	to avoid overrepresentation
	(e.g. above 200 particles per group).

b21.new	
	To make a new selection file using the previous groups.
	
	Output: selec002
	
Check total number of particles

b22.ode	
	Split select file used in 3D into two separate select files to be
        used in the following two 3D reconstructions for comparative 
        purposes.
        p_oddeven.ode	
	output:
	selecodd002 
	seleceven002 

Copy symmetries.xxx file to working directory

b23.bpe	
	Compute the 3D reconstruction of half of the available particles.
        output:
	volume1even

b24.bpo
	Compute the 3D reconstruction of the other half of the available
        particles.
        output:
	volume1odd

b25.res	
	Compare the two volumes (resolution)
	Output:
	rfdoc001

	Resolution value is taken from the third column, when
	it's around 0.5 : first column = 0.24667

	Pixel size = 7.7 (3.85x2, since image was interpolated)

	Resolution = 7.7/  0.24667 = 31.2 A
	
Plot the result in gnuplot
	plot 'rfdoc001.tox' using 3:5 title '11/19/98' with lines

b26.bpr 
	Compute the 3D reconstruction.

Categorize and remove remaining bad particles if necessary
	use categorize from document file
	use doc subtract to remove the bad from the previous selection file

Create stack file with all selected, aligned particles
	b30.stk

b87.pam 	
	3D projection alignment.
	Adapted for 4-fold symmetry.
        Compute a projection of the final volume, calculate distances
        between projections, and convert output to angular doc file.
        Calculate new, refined 3D structure using centered projections
        and the corrected angles from the angular doc file.
        The performance of the procedure (and the quality of the
        reconstruction) can be judged by the average correlation
        coefficient between projected structure and the input data.
        It is stored in the last line (key -1) of the shift files.
        uses procedures:
	Makeselect.pam 
	ali3dapmd4s.pam 
	aln3dapmd4s.pam
	alr3dapmd4s.pam
	combat.pam
	lang4s.pam
	

Get resolution of volumes, e.g.:

	dres002.tox 0.26000   ; 3.85*2=7.7 pixel size / 0.26= 29.6 A resolution

b22.dif	
	Get difference map
	I changed the procedure of Pawel, since his procedure makes
	an arithmetic correction for statistics in case the volumes come
	from different datasets, which is not our case.

	In order to be conservative for the differences, I filter the 
	volume to 32 A (0.24), they are substracted and subsequently 
	the difference to 0.21 (like in PP procedure, 
	where they filter the volume to 0.15 and the difference map to 0.12.

	Output = diff201, volf002, etc