Eiger: Difference between revisions

From XDSwiki
Jump to navigation Jump to search
(hint to latest generate_XDS.INP)
 
(69 intermediate revisions by 3 users not shown)
Line 1: Line 1:
Processing of [https://www.dectris.com/EIGER_X_Features.html Eiger] data is different from processing of conventional data, because the frames are wrapped into [http://www.hdfgroup.org HDF5] files (ending with .h5).
Processing of [https://www.dectris.com/EIGER_X_Features.html Eiger] data is different from processing of conventional data, because the frames are wrapped into [http://www.hdfgroup.org HDF5] files (often ending with .h5). However, with the [[LIB]] feature of XDS and a suitable plugin (preferably [https://github.com/dectris/neggia ''Neggia''], or [https://github.com/DiamondLightSource/durin ''Durin''] for data collected at Diamond Light Source), processing is efficient.


== General aspects ==
== General aspects ==
# It is advisable to use the most recent 64bit version of XDS (since version Oct 15, 2015 the 32bit versions are no longer distributed anyway). The idea of the new framecache in XDS is that RAM is used to save on I/O. To this end, XDS tries to store NUMBER_OF_IMAGES_IN_CACHE=DELPHI/OSCILLATION_RANGE images in memory. Each frame is stored as (number of pixels)*(4 bytes) which means 72 MB in case of the Eiger 16M. As an example: if DELPHI=20 and OSCILLATION_RANGE=0.05 your computer has to have 400*72MB = 29GB of memory (plus some more for the program and the operating system). If it has not, the fallback is to the old behaviour of reading each frame three times.
# The framecache of XDS uses memory to save on I/O; it saves a frame in RAM after reading it for the first time. By default, each XDS (or mcolspot/mintegrate) job stores NUMBER_OF_IMAGES_IN_CACHE=DELPHI/OSCILLATION_RANGE+1 images in memory which corresponds to one DELPHI-sized batch of data. This requires (number of pixels)*(number of jobs)*4 Bytes per frame which amounts to 72 MB in case of the Eiger 16M when running with MAXIMUM_NUBER_OF_JOBS=1. (If DELPHI=20 and OSCILLATION_RANGE=0.05 your computer thus has to have at least 400*72MB = 29GB of memory for each job!). If memory allocation fails, the fallback is to the old behaviour of reading each frame three times (instead of once).
# Dectris provides [https://www.dectris.com/news.html?page=2 H5ToXds] (Linux only!) which is needed by XDS. H5ToXds should be copied to e.g. /usr/local/bin/H5ToXds.bin - note the .bin filename extension! As an alternative, one could use GlobalPhasing's hdf2mini-cbf program (needs autoPROC license) or, from http://www.mrc-lmb.cam.ac.uk/harry/imosflm/ver721/downloads, the eiger2cbf-osx or eiger2cbf-linux program written by T. Nakane (the latter were called miniCBF-Linux and miniCBF-OSX before March 13, 2016).
# Apart from the framecache, XDS needs (number of jobs)*(number of processes)*NX*NY*4 Bytes, plus about one GB for the code.
# For faster processing (Linux only, script needs to be adapted to OSX), the [[Eiger#A_script_for_faster_XDS_processing_of_Eiger_data|shell script]] below should be copied to /usr/local/bin/H5ToXds and made executable (<code>chmod a+rx /usr/local/bin/H5ToXds*</code>). This script ''also'' uses RAM to speed up processing; it uses it for fast storage of the temporary file that H5ToXds/eiger2cbf/hdf2mini-cbf writes, and that each parallel thread ("processor") of XDS reads. The amount of additional RAM this requires is modest (about (number of pixels)*(number of threads) bytes).
# Dectris provides the ''Neggia'' library ([https://github.com/dectris/neggia source],[https://www.dectris.com/support/downloads/sign-in binary]) for native reading of HDF5 files, which can be loaded into XDS at runtime using the <code>[[LIB]]=</code> [http://xds.mpimf-heidelberg.mpg.de/html_doc/xds_parameters.html#LIB= keyword]. With this library (which can also be found at https://{{SERVERNAME}}/pub/linux_bin for Linux, and at https://{{SERVERNAME}}/pub/mac_bin for MacOS), no conversion to CBF or otherwise is necessary. It is therefore just as fast and efficient to read HDF5 files as any other file format. At Diamond Light Source, a different HDF5 format was developed, and this requires the [https://github.com/DiamondLightSource/durin/releases/latest ''Durin'' plugin]. The latter can also read the HDF5 files written by the Dectris software, but frames are not read in parallel, so it is slower.


A suitable [[XDS.INP]] may have been written by the data collection (beamline) software. Latest [[generate_XDS.INP]] or the XDS_from_H5.py script (below) can be used if XDS.INP is not available.
A suitable [[XDS.INP]] may have been written by the data collection (beamline) software. Latest [[generate_XDS.INP]] (<code>generate_XDS.INP xxx_master.h5</code>) or the [[Eiger#Script_for_generating_XDS.INP_from_master.h5|XDS_from_H5.py script]] can be used if XDS.INP is not available.


The number of pixels of the Eiger 16M is three times higher than that of the Pilatus 6M, but since the Eiger firmware update in November 2015, the ("bit shufflle LZ4") compression of the .h5 files containing data is better than that of CBF files, which mostly compensates for the increased number of pixels. However, the size of the *master.h5 file from a Eiger 16M experiment at SLS X06SA is more than 300MB, ''no matter how many frames are collected''. It is therefore advisable to compress (by ~75%) the *master.h5 files on-site, before transferring them home using disk or internet. The fastest (parallel) program with the best compression that I found is [http://lbzip2.org lbzip2] (available from the EPEL repository for RHEL clones). It is supposedly fully compatible with bzip2.
== Compression ==


The number of pixels of the Eiger 16M is three times higher than that of the Pilatus 6M, but since the Eiger firmware update in November 2015, the ("bit shufflle LZ4") compression of the .h5 files containing data is better than that of CBF files, which mostly compensates for the increased number of pixels.


== A benchmark ==
The size of the *master.h5 file from a Eiger 16M experiment at SLS X06SA is more than 300MB, ''no matter how many frames are collected''. It is therefore advisable to compress (by ~75%) the *master.h5 files on-site, before transferring them home using disk or internet. A very fast (parallel) program is [http://lbzip2.org lbzip2] (available from the EPEL repository for RHEL clones). It is supposedly fully compatible with bzip2.


Any comparisons should be based on a common dataset. I downloaded from https://www.dectris.com/datasets.html their latest dataset
Update 2016-06-05 (Toine Schreurs): a HDF5 file may be compressed with [https://www.hdfgroup.org/HDF5/docNewFeatures/FileSpace/h5repack.htm h5repack], ''e.g.'' by <code>h5repack -i <in.h5> -o <out.h5> -f GZIP=6</code> (6 is the default compression level of gzip). This should be a good way to reduce the size of master files while keeping them compatible with processing, but needs to be tested. Whether h5repack uses parallel gzip is not clear from the docs.
ftp://dectris.com/EIGER_16M_Nov2015.tar.bz2 (900 frames) and processed it on a single unloaded CentOS7.2 64bit machine with dual Intel(R) Xeon(R) CPU E5-2667 v2 @ 3.30GHz , HT enabled (showing 32 processors in /proc/cpuinfo), on a local XFS filesystem (all defaults), with four JOBs and 12 PROCESSORS (the XDS.INP that Dectris provides suggests 8 JOBs of 12 PROCESSORS, but I changed that). The numbers below refer to the H5ToXds binary as used in the [[Eiger#A_script_for_faster_XDS_processing_of_Eiger_data|script]] below.


The timing, using the latest XDS (BUILT=20151231), is on the first run
== Troubleshooting ==
INIT:  elapsed wall-clock time      12.0 sec
* make sure that master.h5 and the corresponding data.h5 files remain together as collected, and '''don't rename the data.h5 files''' - they are referred to from master.h5.  If you change the names of the data.h5 files or copy them somewhere else, that link is broken unless you fix master.h5.
COLSPOT: elapsed wall-clock time      44.9 sec
  INTEGRATE: total elapsed wall-clock time      65.1 sec
CORRECT: elapsed wall-clock time        2.9 sec
Total elapsed wall-clock time for XDS      133.6 sec


When I repeat this, I get
= Less efficient way of processing Eiger data, using conversion to CBF =
Total elapsed wall-clock time for XDS      128.3 sec
Repeat once again:
Total elapsed wall-clock time for XDS      129.3 sec
So a bit of cache-warming helps, but not much. This machine has 64GB RAM. From the output of "top", the highest memory usage occurs during INTEGRATE, when each of the mintegrate_par processes consumes up to 7.4% of the memory. In other words, in this way less than 20GB of total memory are used. "top" shows a CPU consumption around (on average) 4 times 650%.


The number of JOBs and PROCESSORs could be optimized. I tried 6 JOBs and get
Since the release of Neggia, a plugin for XDS that parallelizes the reading of images from HDF5 data, conversion by H5ToXds should no longer be required in most usage scenarios. The sections below nevertheless describe this possibility, since preliminary experience with some less common network file systems (apparently GPFS, but not NFS) seems to indicate low performance of Neggia.  
Total elapsed wall-clock time for XDS     120.1 sec
so there's still some room for improvement.


With program versions as of 2016-03-10, eiger2cbf-linux is practically as fast as the H5ToXds binary; hdf2mini-cbf is somewhat slower.
Conversion program options: Dectris provides [https://www.dectris.com/news.html?page=2 H5ToXds] (Linux only!). That program converts (as the name indicates) the HDF5 files to CBF files; however, it does not write the geometry and other information into the CBF header (therefore, [[generate_XDS.INP]] or MOSFLM does not work with these files). Alternatives are GlobalPhasing's hdf2mini-cbf program (does ''not'' need autoPROC license) or, from http://www.mrc-lmb.cam.ac.uk/harry/imosflm/ver721/downloads, the eiger2cbf-osx or eiger2cbf-linux program written by T. Nakane. The latter programs do write a useful CBF header.


When unpacking the .h5 files to .cbf files and processing those, I get on the same machine and with same processing parameters:
H5ToXds and eiger2cbf-osx / eiger2cbf-linux do not work with files produced at Diamond Light Source.  
Total elapsed wall-clock time for XDS      96.3 sec
which indicates a 24% overhead due to the HDF5-to-CBF conversion. However, one has to add to this the time for the HDF5-to-CBF conversion, which is (with 18 parallel H5ToXds jobs each converting 50 frames) 34.2 sec, so overall the "on-the-fly" route using the script below is faster than the "pre-conversion" route, at least on this machine.


On multi-socket machines, there are additional considerations having to do with their NUMA architecture - see [[Performance]].
== A script for faster XDS processing of CBF-converted Eiger data (this is only shown out of historic interest) ==


== Troubleshooting ==
For faster processing, the [[Eiger#A_script_for_faster_XDS_processing_of_CBF-converted Eiger data|shell script]] below should be copied to /usr/local/bin/H5ToXds and made executable (<code>chmod a+rx /usr/local/bin/H5ToXds*</code>). The binary H5ToXds then should be named e.g. /usr/local/bin/H5ToXds.bin - note the .bin filename extension! The script ''also'' uses RAM to speed up processing; it uses it for fast storage of the temporary CBF file that H5ToXds/eiger2cbf/hdf2mini-cbf writes, and that each parallel thread ("processor") of XDS reads. The amount of additional RAM this requires is modest (about (number of pixels)*(number of threads) bytes).
* make sure that master.h5 and the corresponding data.h5 files remain together as collected, and '''don't rename the data.h5 files''' - they are referred to from master.h5.  If you change the names of the data.h5 files or copy them somewhere else, that link is broken unless you fix master.h5. 
* the programs get a lot of testing on RHEL/CentOS/SL. To test if the conversion program work (e.g. on uncommon distros like Mint), run it outside XDS, e.g. <pre> H5ToXds master.h5 1:100 out.cbf </pre> If this creates CBF-compressed files for the first 100 images of your dataset, all is good.


== A script for faster XDS processing of Eiger data ==
<pre>
<pre>
#!/bin/bash
#!/bin/bash
# Kay Diederichs 10/2015
# Kay Diederichs 10/2015
# 3/2016 adapt for eiger2cbf-linux and hdf2min-cbf
# 3/2017 include RAMdisk creation for MacOS; only lightly tested!
# 3/2016 adapt for eiger2cbf and hdf2mini-cbf
# for the latter see https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ccp4bb;58a4ee1.1603 and
# for the latter see https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ccp4bb;58a4ee1.1603 and
# https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ccp4bb;a048b4e8.1603  
# https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ccp4bb;a048b4e8.1603  
Line 61: Line 47:
# Recommendation:
# Recommendation:
# - for the fast local directory one should use a RAMdisk (one GB size at most)
# - for the fast local directory one should use a RAMdisk (one GB size at most)
# - /dev/shm seems to be set up for that purpose on most distributions
# - /dev/shm seems to be already set up for that purpose on most Linux distributions
# - on MacOS you can easily set this up as described at http://stackoverflow.com/questions/2033362/does-os-x-have-an-equivalent-to-dev-shm
# example on MacOS for 1GB RAMdisk (needs to be repeated after booting):
# diskutil eraseVolume HFS+ RAMdisk $(hdiutil attach -nomount ram://$((2 * 1024 * 1000)))
#
#
# on MacOS the next line should then be:
# tempfile="/Volumes/RAMdisk/H5ToXds${PWD//\//_}.$3"
# and on Linux:
tempfile="/dev/shm/H5ToXds${PWD//\//_}.$3"
tempfile="/dev/shm/H5ToXds${PWD//\//_}.$3"
#
#
# choose between H5ToXds.bin,  eiger2cbf and hdf2mini-cbf; un/comment accordingly  
# choose between H5ToXds.bin,  eiger2cbf and hdf2mini-cbf; un/comment accordingly  
/usr/local/bin/H5ToXds.bin $1 $2 "$tempfile"
/usr/local/bin/H5ToXds.bin $1 $2 "$tempfile" || rm "$tempfile"
#/usr/local/bin/eiger2cbf-linux $1 $2 "$tempfile" >& /dev/null
#/usr/local/bin/eiger2cbf-linux $1 $2 "$tempfile" >& /dev/null || rm "$tempfile"
#/usr/local/bin/hdf2mini-cbf $1 $2 "$tempfile"  
#/usr/local/bin/eiger2cbf-osx $1 $2 "$tempfile" >& /dev/null  || rm "$tempfile"
#/usr/local/bin/hdf2mini-cbf $1 $2 "$tempfile"  || rm "$tempfile"
ln -sf "$tempfile" $3 2>/dev/null
ln -sf "$tempfile" $3 2>/dev/null
</pre>
</pre>


== XDS_from_H5.py script for generating XDS.INP given a master .h5 file ==
= See also =
This script could be made executable and put into /usr/local/bin. It requires the [https://www.dectris.com/albula.html#main_head_navigation ALBULA API] to be installed. If you get the error message
ImportError: No module named numpy.core.multiarray
you should
yum -y install numpy
as root.
<pre>
#!/usr/bin/python
# coding: utf8


import sys
[[Performance]]
# Path needs to be be set only if dectris.albula is not found
# i.e. if ALBULA was installed without "--python=</path/to/python_interpreter>"
# Define correct path to ALBULA API:
sys.path.insert(0,"/usr/local/dectris/albula/3.1/python")
try:
  import dectris.albula as dec
except ImportError:
  print "\nThe DECTRIS ALBULA API could not be loaded."
  print "If you did not install ALBULA with \"--python=</path/to/python_interpreter>\","
  print "please modify the \'sys.path.insert\' line in the script to point"
  print "to the DECTRIS ALBULA API and uncomment the line."
  raise SystemExit
 
import os.path
import re
 
# This script was developed by Andreas Förster at DECTRIS based on work by Marcus Mueller.
# Please note that this is not an official DECTRIS product and neither endorsed nor supported by DECTRIS.
# Please report errors and problems to docandreas@gmail.com.
 
XDS_header_lines = """!*****************************************************************************
!
! XDS.INP template for ! %(family)s %(detector)s with %(sensor).2f mm thick silicon sensors.
!
!    Characters to the right of an exclamation mark are comments.
!
!    This file was autogenerated by XDS_from_H5.py (Oct 2015).
!    Please check default values before processing.
!
! For questions and comments please contact docandreas@gmail.com.
!
!*****************************************************************************
"""
 
XDS_detector_lines = """
!====================== DETECTOR PARAMETERS ==================================
DETECTOR=%(family)s
MINIMUM_VALID_PIXEL_VALUE=0
OVERLOAD= %(cutoff)i ! taken from HDF5 header item
! /entry/instrument/detector/detectorSpecific/countrate_correction_count_cutoff
SENSOR_THICKNESS=%(sensor).2f ! [mm]
!SILICON=-1.0
QX=%(pixsize_x).3f  QY=%(pixsize_y).3f  ! [mm]
"""
 
XDS_main_lines = """
TRUSTED_REGION=0.0 1.41 !Relative radii limiting trusted detector region
 
DIRECTION_OF_DETECTOR_X-AXIS= 1.0 0.0 0.0
DIRECTION_OF_DETECTOR_Y-AXIS= 0.0 1.0 0.0 ! 0.0 cos(2theta) sin(2theta)
 
!====================== JOB CONTROL PARAMETERS ===============================
JOB= XYCORR INIT COLSPOT IDXREF DEFPIX ! XPLAN INTEGRATE CORRECT
 
MAXIMUM_NUMBER_OF_JOBS=8  !Speeds up COLSPOT & INTEGRATE on multicore machine
MAXIMUM_NUMBER_OF_PROCESSORS=4!<32;ignored by single cpu version of xds
!SECONDS=0  !Maximum number of seconds to wait until data image must appear
!TEST=1    !Test flag. 1,2 additional diagnostics and images
 
!====================== GEOMETRICAL PARAMETERS ===============================
!ORGX and ORGY are often close to the image center, i.e. ORGX=NX/2, ORGY=NY/2
ORGX= %(orgx).1f  ORGY= %(orgy).1f    !Detector origin (pixels).  ORGX=NX/2; ORGY=NY/2
DETECTOR_DISTANCE= %(dist)i  ! [mm]
 
ROTATION_AXIS= 1.0 0.0 0.0
 
! Optimal choice is 0.5*mosaicity (REFLECTING_RANGE_E.S.D.= mosaicity)
OSCILLATION_RANGE=%(osc_range).5f            ! [deg] (>0)
 
X-RAY_WAVELENGTH=%(wavelength).4f          ! [A]
INCIDENT_BEAM_DIRECTION=0.0 0.0 1.0
FRACTION_OF_POLARIZATION=0.99 !default=0.5 for unpolarized beam
POLARIZATION_PLANE_NORMAL= 0.0 1.0 0.0
 
!======================= CRYSTAL PARAMETERS =================================
SPACE_GROUP_NUMBER=0  !0 for unknown crystals; cell constants are ignored.
UNIT_CELL_CONSTANTS= 0 0 0 0 0 0
 
! You may specify here the x,y,z components for the unit cell vectors if
! known from a previous run using the same crystal in the same orientation
!UNIT_CELL_A-AXIS=
!UNIT_CELL_B-AXIS=
!UNIT_CELL_C-AXIS=
 
!Optional reindexing transformation to apply on reflection indices
!REIDX=  0  0 -1  0  0 -1  0  0 -1  0  0  0
 
FRIEDEL'S_LAW=FALSE ! Default is TRUE.
 
!REFERENCE_DATA_SET= CK.HKL  ! Name of a reference data set (optional)
 
!==================== SELECTION OF DATA IMAGES ==============================
!Generic file name and format (optional) of data images
NAME_TEMPLATE_OF_DATA_FRAMES=%(name_template)s ! HDF5
"""
 
XDS_tail_lines = """
!==================== DATA COLLECTION STRATEGY (XPLAN) ======================
!                      !!! Warning !!!
! If you processed your data for a crystal with unknown cell constants and
! space group symmetry, XPLAN will report the results for space group P1.
 
!STARTING_ANGLE=  0.0      STARTING_FRAME=1
!used to define the angular origin about the rotation axis.
!Default:  STARTING_ANGLE=  0 at STARTING_FRAME=first data image
 
!RESOLUTION_SHELLS=10 6 5 4 3 2 1.5 1.3 1.2
 
!STARTING_ANGLES_OF_SPINDLE_ROTATION= 0 180 10
 
!TOTAL_SPINDLE_ROTATION_RANGES=30.0 120 15
 
!====================== INDEXING PARAMETERS =================================
!Never forget to check this, since the default 0 0 0 is almost always correct!
!INDEX_ORIGIN= 0 0 0          ! used by "IDXREF" to add an index offset
 
!Additional parameters for fine tuning that rarely need to be changed
!INDEX_ERROR=0.05 INDEX_MAGNITUDE=8 INDEX_QUALITY=0.8
SEPMIN=4.0      ! default is 6 for other detectors
CLUSTER_RADIUS=2 ! default is 3 for other detectors
!MAXIMUM_ERROR_OF_SPOT_POSITION=3.0
!MAXIMUM_ERROR_OF_SPINDLE_POSITION=2.0
!MINIMUM_FRACTION_OF_INDEXED_SPOTS=0.5
 
!============== DECISION CONSTANTS FOR FINDING CRYSTAL SYMMETRY =============
!Decision constants for detection of lattice symmetry (IDXREF, CORRECT)
MAX_CELL_AXIS_ERROR= 0.03 ! Maximum relative error in cell axes tolerated
MAX_CELL_ANGLE_ERROR= 2.0 ! Maximum cell angle error tolerated
 
!Decision constants for detection of space group symmetry (CORRECT).
!Resolution range for accepting reflections for space group determination in
!the CORRECT step. It should cover a sufficient number of strong reflections.
TEST_RESOLUTION_RANGE= 8.0 4.5
MIN_RFL_Rmeas= 50  ! Minimum #reflections needed for calculation of Rmeas
MAX_FAC_Rmeas= 2.0 ! Sets an upper limit for acceptable Rmeas
 
!================= PARAMETERS CONTROLLING REFINEMENTS =======================
REFINE(IDXREF)= BEAM AXIS ORIENTATION CELL ! POSITION
REFINE(INTEGRATE)= POSITION ORIENTATION ! BEAM CELL AXIS
REFINE(CORRECT)= POSITION BEAM ORIENTATION CELL AXIS
 
!================== CRITERIA FOR ACCEPTING REFLECTIONS ======================
VALUE_RANGE_FOR_TRUSTED_DETECTOR_PIXELS= 6000 30000 !Used by DEFPIX
                  !for excluding shaded parts of the detector.
 
INCLUDE_RESOLUTION_RANGE=50.0 %(reso_range).1f !Angstroem; used by DEFPIX,INTEGRATE,CORRECT
 
!used by CORRECT to exclude ice-reflections
!EXCLUDE_RESOLUTION_RANGE= 3.93 3.87 !ice-ring at 3.897 Angstrom
!EXCLUDE_RESOLUTION_RANGE= 3.70 3.64 !ice-ring at 3.669 Angstrom
!EXCLUDE_RESOLUTION_RANGE= 3.47 3.41 !ice-ring at 3.441 Angstrom
!EXCLUDE_RESOLUTION_RANGE= 2.70 2.64 !ice-ring at 2.671 Angstrom
!EXCLUDE_RESOLUTION_RANGE= 2.28 2.22 !ice-ring at 2.249 Angstrom
!EXCLUDE_RESOLUTION_RANGE= 2.102 2.042 !ice-ring at 2.072 Angstrom - strong
!EXCLUDE_RESOLUTION_RANGE= 1.978 1.918 !ice-ring at 1.948 Angstrom - weak
!EXCLUDE_RESOLUTION_RANGE= 1.948 1.888 !ice-ring at 1.918 Angstrom - strong
!EXCLUDE_RESOLUTION_RANGE= 1.913 1.853 !ice-ring at 1.883 Angstrom - weak
!EXCLUDE_RESOLUTION_RANGE= 1.751 1.691 !ice-ring at 1.721 Angstrom - weak
 
!MINIMUM_ZETA=0.05 !Defines width of 'blind region' (XPLAN,INTEGRATE,CORRECT)
 
!WFAC1=1.0  !This controls the number of rejected MISFITS in CORRECT;
        !a larger value leads to fewer rejections.
!REJECT_ALIEN=20.0 ! Automatic rejection of very strong reflections
 
!============== INTEGRATION AND PEAK PROFILE PARAMETERS =====================
!Specification of the peak profile parameters below overrides the automatic
!determination from the images
!Suggested values are listed near the end of INTEGRATE.LP
!BEAM_DIVERGENCE=  0.80        !arctan(spot diameter/DETECTOR_DISTANCE)
!BEAM_DIVERGENCE_E.S.D.=  0.080 !half-width (Sigma) of BEAM_DIVERGENCE
!REFLECTING_RANGE=  0.780 !for crossing the Ewald sphere on shortest route
!REFLECTING_RANGE_E.S.D.=  0.113 !half-width (mosaicity) of REFLECTING_RANGE
 
NUMBER_OF_PROFILE_GRID_POINTS_ALONG_ALPHA/BETA=21!used by: INTEGRATE
NUMBER_OF_PROFILE_GRID_POINTS_ALONG_GAMMA=21    !used by: INTEGRATE
 
!DELPHI= 6.0!controls the number of reference profiles and scaling factors
!CUT=2.0    !defines the integration region for profile fitting
!MINPK=75.0 !minimum required percentage of observed reflection intensity
 
!======= PARAMETERS CONTROLLING CORRECTION FACTORS (used by: CORRECT) =======
!MINIMUM_I/SIGMA=3.0 !minimum intensity/sigma required for scaling reflections
!NBATCH=-1  !controls the number of correction factors along image numbers
!REFLECTIONS/CORRECTION_FACTOR=50  !minimum #reflections/correction needed
!PATCH_SHUTTER_PROBLEM=TRUE        !FALSE is default
!STRICT_ABSORPTION_CORRECTION=TRUE  !FALSE is default
!CORRECTIONS= DECAY MODULATION ABSORPTION
 
!=========== PARAMETERS DEFINING BACKGROUND AND PEAK PIXELS =================
!STRONG_PIXEL=3.0                              !used by: COLSPOT
!A 'strong' pixel to be included in a spot must exceed the background
!by more than the given multiple of standard deviations.
 
!MAXIMUM_NUMBER_OF_STRONG_PIXELS=1500000      !used by: COLSPOT
 
!SPOT_MAXIMUM-CENTROID=3.0                    !used by: COLSPOT
 
MINIMUM_NUMBER_OF_PIXELS_IN_A_SPOT=3          !used by: COLSPOT
!This allows to suppress spurious isolated pixels from entering the
!spot list generated by "COLSPOT".
 
!NBX=3  NBY=3  !Define a rectangle of size (2*NBX+1)*(2*NBY+1)
!The variation of counts within the rectangle centered at each image pixel
!is used for distinguishing between background and spot pixels.
 
!BACKGROUND_PIXEL=6.0                          !used by: COLSPOT,INTEGRATE
!An image pixel does not belong to the background region if the local
!pixel variation exceeds the expected variation by the given number of
!standard deviations.
 
!SIGNAL_PIXEL=3.0                              !used by: INTEGRATE
!A pixel above the threshold contributes to the spot centroid
 
!FIXED_SCALE_FACTOR=TRUE  !Default is FALSE; used by : INIT,INTEGRATE
"""
 
detector_families = {
    'pilatus' : {
        'nmodules' : {
            '12M': (5, 24),
            '6M' : (5, 12),
            '2M' : (3 ,8),
            '1M' : (2, 5),
            '300K-W': (3, 1),
            '300K' : (1 ,3),
            '200K' : (1 ,2),
            '100K' : (1, 1),
            },
        'module' : {
            'size': (487, 195),
            'gap': (7, 17),
            'pixel_size': (0.172e-03, 0.172e-03),
            'nchips': (8, 2),
            },
        'chip': {
            'size': (60, 97),
            'gap': (1, 1),
        },
        'sizes' : {}, # will be populated with correct sizes
        },
    'eiger' : {
        'nmodules' : {
            '1M': (1, 2),
            '4M': (2, 4),
            '9M': (3, 6),
            '16M': (4, 8),
            },
        'module' : {
            'size': (1030, 514),
            'gap': (10, 37),
            'pixel_size': (0.075e-03, 0.075e-03),
            'nchips': (4, 2),
            },
        'chip' : {
            'size' : (256, 256),
            'gap' : (2, 2),
        },
        'sizes' : {}, # will be populated with correct sizes
        },
    }
 
# All interesting parameters
incident_wavelength = "/entry/instrument/beam/incident_wavelength"
software_version = "/entry/instrument/detector/detectorSpecific/software_version"
beam_center_x = "/entry/instrument/detector/beam_center_x"
beam_center_y = "/entry/instrument/detector/beam_center_y"
x_pixel_size = "/entry/instrument/detector/x_pixel_size"
y_pixel_size = "/entry/instrument/detector/y_pixel_size"
detector_distance = "/entry/instrument/detector/detector_distance"
sensor_thickness = "/entry/instrument/detector/sensor_thickness"
nimages = "/entry/instrument/detector/detectorSpecific/nimages"
description = "/entry/instrument/detector/description"
omega_range_average = "/entry/sample/goniometer/omega_range_average"
countrate_correction_count_cutoff = "/entry/instrument/detector/detectorSpecific/countrate_correction_count_cutoff"
resolution_cutoff = 'max resolution'
 
# The list below contains the parameters to be extracted from H5
parameters = [
    incident_wavelength,
    software_version,
    beam_center_x,
    beam_center_y,
    x_pixel_size,
    y_pixel_size,
    detector_distance,
    sensor_thickness,
    nimages,
    description,
    omega_range_average,
    countrate_correction_count_cutoff,
    resolution_cutoff
    ]
 
def create_XDS_INP(parameters, file_name):
    lines = []
    description = parameters["/entry/instrument/detector/description"].split()
    family = description[1].lower()
    sensor = float(parameters["/entry/instrument/detector/sensor_thickness"]) * 1000.0
    FAMILY = family.upper()
    det_name = description[2]
    file_template = re.sub("master\.h5", "??????.h5", file_name)
    lines.append(XDS_header_lines % {
        'family': FAMILY,
        'detector': det_name,
        'sensor': sensor,})
    lines.append(XDS_detector_lines % {
        'family': FAMILY,
        'cutoff': int(float(parameters["/entry/instrument/detector/detectorSpecific/countrate_correction_count_cutoff"])),
        'sensor': sensor,
        'pixsize_x': float(parameters["/entry/instrument/detector/x_pixel_size"]) * 1000.0,
        'pixsize_y': float(parameters["/entry/instrument/detector/y_pixel_size"]) * 1000.0,})
    lines = lines + get_size_specific_lines(fam=family, det=det_name, n_excluded_edge_pixels=0)
    lines.append(XDS_main_lines % {
        'orgx': float(parameters["/entry/instrument/detector/beam_center_x"]),
        'orgy': float(parameters["/entry/instrument/detector/beam_center_y"]),
        'dist': int(float(parameters["/entry/instrument/detector/detector_distance"]) * 1000.0),
        'osc_range': float(parameters["/entry/sample/goniometer/omega_range_average"]),
        'wavelength': float(parameters["/entry/instrument/beam/incident_wavelength"]),
        'name_template': file_template,})
    first = 1
    last = int(parameters["/entry/instrument/detector/detectorSpecific/nimages"])
    para_images = int(full_parameters["/entry/instrument/detector/detectorSpecific/nimages"])
    rotation = float(full_parameters["/entry/sample/goniometer/omega_range_average"])
    lines.append("\n DATA_RANGE=%i %i\n" % (first, last))
    if (para_images * rotation <= 30):
        if (last > 100):
            bkg = 100
        else:
            bkg = last
        lines.append("\n")
        lines.append(" BACKGROUND_RANGE=%i %i  ! Numbers of first and last data image for background\n" % (first, bkg))
        lines.append("!Five degrees are sufficient\n")
        lines.append("\n")
        lines.append(" SPOT_RANGE= %i %i      ! Image range for finding spots\n" % (first, last))
        lines.append("!Use all images if this range is not sufficient\n")
    elif (para_images * rotation > 30):
        # split spot finding into three 10 degree segments
        bkg = first + int(5/rotation)
        end1  = first + int(10/rotation)
        start2 = first + int(last/2)
        end2  = first + int(last/2) + int(10/rotation)
        start3 = first + last - int(10/rotation) - 1
        end3  = first + last - 1
        lines.append("\n")
        lines.append(" BACKGROUND_RANGE=%i %i  ! Numbers of first and last data image for background\n" % (first, bkg))
        lines.append("!Five degrees are sufficient\n")
        lines.append("\n")
        lines.append(" SPOT_RANGE= %i %i      ! First image range for finding spots\n" % (first, end1))
        lines.append(" SPOT_RANGE= %i %i      ! Second image range for finding spots\n" % (start2, end2))
        lines.append(" SPOT_RANGE= %i %i      ! Third image range for finding spots\n" % (start3, end3))
        lines.append("!Use all images if three ranges are not sufficient\n")
    lines.append(XDS_tail_lines % {
        'reso_range': float(parameters["max resolution"]),})
    return lines
 
def get_size_specific_lines(fam, det, n_excluded_edge_pixels=0):
    param_lines = []
    gaps = calculate_gaps(
        detector_families[fam]['sizes'][det],
        detector_families[fam]['module']['size'],
        detector_families[fam]['module']['gap'],
        )
    param_lines.append(' NX= %4d  NY= %4d \n\n' % detector_families[fam]['sizes'][det])
    param_lines.append('!EXCLUSION OF VERTICAL DEAD AREAS OF THE '
        '%s %s DETECTOR \n' % (fam.upper(), det))
    module_edge_comment = ('!EXCLUDING %d ADDITIONAL PIXELS OF THE '
        'MODULE EDGES \n' % n_excluded_edge_pixels)
    if n_excluded_edge_pixels > 0:
        param_lines.append(module_edge_comment)
    # offset is required because XDS.INP pixel values start with 1, not 0
    offset = 1
    for gap in gaps[0]:
        param_lines.append(' UNTRUSTED_RECTANGLE= %4d %4d  %4d %4d \n' % (
            gap[0] - 1 + offset - n_excluded_edge_pixels,
            gap[1] + 1 + offset + n_excluded_edge_pixels,
            0,
            detector_families[fam]['sizes'][det][1] + 1))
    param_lines.append('\n')
    param_lines.append('!EXCLUSION OF HORIZONTAL DEAD AREAS OF THE '
        '%s %s DETECTOR \n' % (fam.upper(), det))
    if n_excluded_edge_pixels > 0:
        param_lines.append(module_edge_comment)
    for gap in gaps[1]:
        param_lines.append(' UNTRUSTED_RECTANGLE= %4d %4d  %4d %4d \n' % (
            0,
            detector_families[fam]['sizes'][det][0] + 1,
            gap[0] - 1 + offset - n_excluded_edge_pixels,
            gap[1] + 1 + offset + n_excluded_edge_pixels))
    return param_lines
 
def warning():
    return ('\nThis script extracts from a given HDF5 master file all metadata\n'
            'required to write XDS.INP.  The user is prompted for missing metadata.\n'
            '\n'
            'WARNING - This script is a proof-of-principle, pre-alpha.\n'
            'Do not rely on it for anything serious.  Things will go wrong.\n'
            'In particular, this does not work for data collected in ROI mode.\n'
            '\n'
            'Please report shortcomings and errors to andreas.foerster@dectris.com\n')
 
def help():
    return ('ERROR - You must specify exactly one HDF5 master file:\n'
            '\n'
            'python XDS_from_H5.py <name>_master.h5\n')
 
def version_check(version):
    if float(re.search('^\d+\.\d+', version).group(0)) > 1.2:
        return 1
    else:
        return 0
 
zero_values = [0, "0", 0.0, "0.0"]
 
def isFile(file_input):
    '''This function verifies that the file name entered by the user
    corresponds to a master.h5 file and attaches an extension if necessary.'''
    if os.path.isfile(file_input) and re.search("master\.h5\Z", file_input):
        return file_input
    elif os.path.isfile(file_input + ".h5") and re.search("master\Z", file_input):
        return(file_input + ".h5")
    else:
        return 0
 
def request_parameter(parameter):
    if (parameter == omega_range_average):
        return raw_input("Please enter the oscillation range in degrees.\n")
    elif (parameter == detector_distance):
        return raw_input("Please enter the detector distance in meters.\n")
    elif (parameter == incident_wavelength):
        return raw_input("Please enter the wavelength in Angstrom.\n")
    elif (parameter == beam_center_x):
        return raw_input("Please enter the x coordinate of the beam center in pixels.\n")
    elif (parameter == beam_center_y):
        return raw_input("Please enter the y coordinate of the beam center in pixels.\n")
    elif (parameter == x_pixel_size):
        return raw_input("Please enter the x coordinate of the pixel size.\n")
    elif (parameter == y_pixel_size):
        return raw_input("Please enter the y coordinate of the pixel size.\n")
    elif (parameter == sensor_thickness):
        return raw_input("Please enter the sensor thickness in meters.\n")
    elif (parameter == nimages):
        return raw_input("Please enter the number of images.\n")
    elif (parameter == description):
        print "Please enter the description of the detector, e.g."
        return raw_input("Dectris Eiger 4M\n")
    elif (parameter == countrate_correction_count_cutoff):
        return raw_input("Please enter the maximum trusted pixel value.\n")
    elif (parameter == resolution_cutoff):
        return raw_input("Please enter a resolution limit for processing.\n")
    else:
        print "Unknown software version.  Please check."
        return 0
 
def calculate_gaps(det_size, mod_size, gap_size):
    """
    Return list of tuples with first and last pixel in each detector gap.
 
    One list for each detector dimension (x and y).
    Input: total detector size in pixels
        size of a module in pixels
        size of a gap in pixels
    """
    ndims = len(det_size)
    gaps = []
    for dim_index in range(ndims):
        gaps.append([])
        module_start = 0
        while module_start < det_size[dim_index]:
            # First pixel on a module has index 0, Python and C style
            gap_start = module_start + mod_size[dim_index]
            module_start = gap_start + gap_size[dim_index]
            gap_end = module_start - 1
            if module_start < det_size[dim_index]:
                gaps[dim_index].append((gap_start, gap_end))
            else:
                break
    return gaps
 
 
 
# Creates a dictionary of all keys and values in the NeXus tree
def iterate_children(node, nodeDict={}):
    """ iterate over the children of a neXus node """
    if node.type() == dec.DNeXusNode.GROUP:
        for kid in node.children():
            nodeDict = iterate_children(kid, nodeDict)
    else:
        nodeDict[node.path()] = node.value()
    return nodeDict


# Extracts values from HDF5 file according to parameters array
[https://github.com/keitaroyam/yamtbx/blob/master/doc/eiger-en.md Keitaro Yamashita's Eiger page, with some emphasis on SPring-8]
def get_params(hdf5_file):
    extracted = {}
    h5cont = dec.DImageSeries(hdf5_file)
    neXus_tree = h5cont.neXus()
    neXus_root = neXus_tree.root()
    neXus_string_tree = iterate_children(neXus_root)
    if (len(sys.argv) == 2):
        print "Extracting metadata from " + hdf5_file
        print "Please modify XDS.INP if these numbers are incorrect.\n"
    for i in parameters:
        if (neXus_string_tree.has_key(i)):
            extracted[i] = str(neXus_string_tree[i])
        else:
            extracted[i] = ""
    return extracted
 
def calculate_size(n_modules, mod_size, gap_size):
    n_gaps = [n - 1 for n in n_modules]
    size = []
    for nmod, ngap, nmodpix, ngappix in zip(n_modules, n_gaps, mod_size, gap_size):
        size.append(nmod * nmodpix + ngap * ngappix)
    return tuple(size)
 
# populate dicts with sizes of detectors in pixels
for family in detector_families.values():
    for model, n_modules in family['nmodules'].items():
        family['sizes'][model] = calculate_size(n_modules=n_modules,
                mod_size=family['module']['size'],
                gap_size=family['module']['gap'])
 
if __name__ == "__main__":
    if len(sys.argv) == 2:
        # Make sure that XDS.INP does not already exist
        if os.path.isfile ("XDS.INP"):
            print "\nERROR: XDS.INP exists already.  Please rename and rerun script."
        else:
            # test whether argument 1 is HDF5 file.
            # attach ".h5" if necessary
            clean_file = isFile(sys.argv[1])
            if (clean_file):
                print warning()
                full_parameters = get_params(clean_file)
                for i, v in full_parameters.iteritems():
                    if (v in zero_values):
                        print i + " = " + str(v) + "  <== WARNING:  Should this really be 0?"
                        full_parameters[i] = request_parameter(i)
                        print i + " = " + str(full_parameters[i])
                    elif (v == "NaN") or (v == ""):
                        print i + " = " + v + "  <== ERROR:  undefined value."
                        full_parameters[i] = request_parameter(i)
                        print i + " = " + str(full_parameters[i])
                    else:
                        print i + " = " + str(v)
                para_version = str(full_parameters[software_version])
                if version_check(para_version):
                    param_lines = create_XDS_INP(full_parameters, clean_file)
                    open("XDS.INP", 'w').writelines(param_lines)
                    print "\nFile XDS.INP was created successfully."
                    if (int(full_parameters["/entry/instrument/detector/detectorSpecific/nimages"]) == 1):
                        print "However, there's not much you can do with one image.\n"
                    else:
                        print "Please verify its contents before processing data.\n"
                else:
                    print "\nThe HDF5 file was created with version %s of the detector firmware" % (para_version)
                    print "This script supports versions 1.2 and up."
                    print "\nFile XDS.INP was not created."
                    print "Please extract metadata with hdfview or h5dump.\n"
            else:
                print help()
    elif (len(sys.argv) == 3):
        # This assumes the second argument is the rotation range
        # The script will run non-interactively
        # The master.h5 must be specified with its full name
        # An existing XDS.INP will be overwritten
        full_parameters = get_params(sys.argv[1])
        full_parameters["omega_range_average"] = sys.argv[2]
        param_lines = create_XDS_INP(full_parameters, sys.argv[1])
        open("XDS.INP", 'w').writelines(param_lines)
    else:
        print help()
        exit(-1)
</pre>
 
 
== See also ==
 
[[Performance]]

Latest revision as of 23:07, 16 August 2022

Processing of Eiger data is different from processing of conventional data, because the frames are wrapped into HDF5 files (often ending with .h5). However, with the LIB feature of XDS and a suitable plugin (preferably Neggia, or Durin for data collected at Diamond Light Source), processing is efficient.

General aspects

  1. The framecache of XDS uses memory to save on I/O; it saves a frame in RAM after reading it for the first time. By default, each XDS (or mcolspot/mintegrate) job stores NUMBER_OF_IMAGES_IN_CACHE=DELPHI/OSCILLATION_RANGE+1 images in memory which corresponds to one DELPHI-sized batch of data. This requires (number of pixels)*(number of jobs)*4 Bytes per frame which amounts to 72 MB in case of the Eiger 16M when running with MAXIMUM_NUBER_OF_JOBS=1. (If DELPHI=20 and OSCILLATION_RANGE=0.05 your computer thus has to have at least 400*72MB = 29GB of memory for each job!). If memory allocation fails, the fallback is to the old behaviour of reading each frame three times (instead of once).
  2. Apart from the framecache, XDS needs (number of jobs)*(number of processes)*NX*NY*4 Bytes, plus about one GB for the code.
  3. Dectris provides the Neggia library (source,binary) for native reading of HDF5 files, which can be loaded into XDS at runtime using the LIB= keyword. With this library (which can also be found at https://wiki.uni-konstanz.de/pub/linux_bin for Linux, and at https://wiki.uni-konstanz.de/pub/mac_bin for MacOS), no conversion to CBF or otherwise is necessary. It is therefore just as fast and efficient to read HDF5 files as any other file format. At Diamond Light Source, a different HDF5 format was developed, and this requires the Durin plugin. The latter can also read the HDF5 files written by the Dectris software, but frames are not read in parallel, so it is slower.

A suitable XDS.INP may have been written by the data collection (beamline) software. Latest generate_XDS.INP (generate_XDS.INP xxx_master.h5) or the XDS_from_H5.py script can be used if XDS.INP is not available.

Compression

The number of pixels of the Eiger 16M is three times higher than that of the Pilatus 6M, but since the Eiger firmware update in November 2015, the ("bit shufflle LZ4") compression of the .h5 files containing data is better than that of CBF files, which mostly compensates for the increased number of pixels.

The size of the *master.h5 file from a Eiger 16M experiment at SLS X06SA is more than 300MB, no matter how many frames are collected. It is therefore advisable to compress (by ~75%) the *master.h5 files on-site, before transferring them home using disk or internet. A very fast (parallel) program is lbzip2 (available from the EPEL repository for RHEL clones). It is supposedly fully compatible with bzip2.

Update 2016-06-05 (Toine Schreurs): a HDF5 file may be compressed with h5repack, e.g. by h5repack -i <in.h5> -o <out.h5> -f GZIP=6 (6 is the default compression level of gzip). This should be a good way to reduce the size of master files while keeping them compatible with processing, but needs to be tested. Whether h5repack uses parallel gzip is not clear from the docs.

Troubleshooting

  • make sure that master.h5 and the corresponding data.h5 files remain together as collected, and don't rename the data.h5 files - they are referred to from master.h5. If you change the names of the data.h5 files or copy them somewhere else, that link is broken unless you fix master.h5.

Less efficient way of processing Eiger data, using conversion to CBF

Since the release of Neggia, a plugin for XDS that parallelizes the reading of images from HDF5 data, conversion by H5ToXds should no longer be required in most usage scenarios. The sections below nevertheless describe this possibility, since preliminary experience with some less common network file systems (apparently GPFS, but not NFS) seems to indicate low performance of Neggia.

Conversion program options: Dectris provides H5ToXds (Linux only!). That program converts (as the name indicates) the HDF5 files to CBF files; however, it does not write the geometry and other information into the CBF header (therefore, generate_XDS.INP or MOSFLM does not work with these files). Alternatives are GlobalPhasing's hdf2mini-cbf program (does not need autoPROC license) or, from http://www.mrc-lmb.cam.ac.uk/harry/imosflm/ver721/downloads, the eiger2cbf-osx or eiger2cbf-linux program written by T. Nakane. The latter programs do write a useful CBF header.

H5ToXds and eiger2cbf-osx / eiger2cbf-linux do not work with files produced at Diamond Light Source.

A script for faster XDS processing of CBF-converted Eiger data (this is only shown out of historic interest)

For faster processing, the shell script below should be copied to /usr/local/bin/H5ToXds and made executable (chmod a+rx /usr/local/bin/H5ToXds*). The binary H5ToXds then should be named e.g. /usr/local/bin/H5ToXds.bin - note the .bin filename extension! The script also uses RAM to speed up processing; it uses it for fast storage of the temporary CBF file that H5ToXds/eiger2cbf/hdf2mini-cbf writes, and that each parallel thread ("processor") of XDS reads. The amount of additional RAM this requires is modest (about (number of pixels)*(number of threads) bytes). 

#!/bin/bash
# Kay Diederichs 10/2015
# 3/2017 include RAMdisk creation for MacOS; only lightly tested!
# 3/2016 adapt for eiger2cbf and hdf2mini-cbf
# for the latter see https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ccp4bb;58a4ee1.1603 and
# https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ccp4bb;a048b4e8.1603 
#
# Idea: put temporary files into fast local directory, instead of NFS
#
# Installation: Rename Dectris' H5ToXds to H5ToXds.bin
#               This script should be called H5ToXds and reside in $PATH 
#               Modify this script according to which binary you use - see comments below.
#
# Recommendation:
# - for the fast local directory one should use a RAMdisk (one GB size at most)
# - /dev/shm seems to be already set up for that purpose on most Linux distributions
# - on MacOS you can easily set this up as described at http://stackoverflow.com/questions/2033362/does-os-x-have-an-equivalent-to-dev-shm
# example on MacOS for 1GB RAMdisk (needs to be repeated after booting):
# diskutil eraseVolume HFS+ RAMdisk $(hdiutil attach -nomount ram://$((2 * 1024 * 1000)))
#
# on MacOS the next line should then be:
# tempfile="/Volumes/RAMdisk/H5ToXds${PWD//\//_}.$3"
# and on Linux:
tempfile="/dev/shm/H5ToXds${PWD//\//_}.$3"
#
# choose between H5ToXds.bin,  eiger2cbf and hdf2mini-cbf; un/comment accordingly 
/usr/local/bin/H5ToXds.bin $1 $2 "$tempfile" || rm "$tempfile"
#/usr/local/bin/eiger2cbf-linux $1 $2 "$tempfile" >& /dev/null  || rm "$tempfile"
#/usr/local/bin/eiger2cbf-osx $1 $2 "$tempfile" >& /dev/null  || rm "$tempfile"
#/usr/local/bin/hdf2mini-cbf $1 $2 "$tempfile"  || rm "$tempfile"
ln -sf "$tempfile" $3 2>/dev/null

See also

Performance

Keitaro Yamashita's Eiger page, with some emphasis on SPring-8

Retrieved from "https://wiki.uni-konstanz.de/xds/index.php?title=Eiger&oldid=4441"