Pair of same helical segments found after symmetry expansion (RELION-3.1-beta-commit-a0af57)

[acook2]

Problem Description Particle symmetry expansion was performed on a helical dataset. After this, 3D auto-refine was performed, which crashed after the maximisation step of the first iteration. Presumed to be because around line 4034 of helix.cpp, particles are tested for equivalence of their rlnHelicalTrackLength value - but after symmetry expansion, the expanded particles have the same rlnHelicalTrackLength as their parent particle. This idea was tested with tubes where the rlnHelicalTrackLength of expanded particles had been edited to be non-identical, which stopped the error.

Environment:

• CentOS release 6.10
• RELION-3.1-beta-commit-a0af57

Dataset:

• Box size: 432
• Pixel size: 1.39
• Description: Helix (rise = 8.8 Å, twist = -25.7 o)

Job options:

• Type of job: Refine3D
• Number of MPI processes: 5
• Number of threads: 3
• Full command:

  `which relion_refine_mpi` --o Refine3D/job087/run --auto_refine --split_random_halves --i Extract/job077/temp.star --ref ../references/dcx_gdp_bx432_px1-39.mrc --firstiter_cc --ini_high 15 --dont_combine_weights_via_disc --pool 100 --pad 2  --skip_gridding  --ctf --ctf_corrected_ref --particle_diameter 480 --flatten_solvent --zero_mask --solvent_mask MaskCreate/job033/mask.mrc --solvent_correct_fsc  --oversampling 1 --healpix_order 5 --auto_local_healpix_order 6 --offset_range 8 --offset_step 2 --sym C1 --low_resol_join_halves 40 --norm --scale  --helix --helical_inner_diameter 190 --helical_outer_diameter 350 --ignore_helical_symmetry --sigma_tilt 1.66667 --sigma_psi 1 --sigma_rot 1.66667 --helical_sigma_distance 1 --helical_keep_tilt_prior_fixed --j 3 --gpu ""  --pipeline_control Refine3D/job087/

Error message:

Compare # 1 with # 0
in: /s/emib/s/relion/v3.1/relion/src/helix.cpp, line 4070
ERROR: 
helix.h::HelicalSegmentPriorInfoEntry::operator<(): A pair of same helical segments is found!
=== Backtrace  ===
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi(_ZN11RelionErrorC1ERKSsS1_l+0x41) [0x49ab71]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi(_ZNK28HelicalSegmentPriorInfoEntryltERKS_+0x223) [0x4a0923]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi(_ZSt16__insertion_sortIN9__gnu_cxx17__normal_iteratorIP28HelicalSegmentPriorInfoEntrySt6vectorIS2_SaIS2_EEEENS0_5__ops15_Iter_less_iterEEvT_SA_T0_+0x3b) [0x4d83eb]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi() [0x4d8e3f]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi(_ZSt22__stable_sort_adaptiveIN9__gnu_cxx17__normal_iteratorIP28HelicalSegmentPriorInfoEntrySt6vectorIS2_SaIS2_EEEES3_lNS0_5__ops15_Iter_less_iterEEvT_SA_T0_T1_T2_+0x5f) [0x4dd32f]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi(_Z36updatePriorsForHelicalReconstructionR13MetaDataTableRiS1_dSt6vectorIdSaIdEES4_bbbddddbd+0xc30) [0x4d00d0]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi(_ZN14MlOptimiserMpi7iterateEv+0xa7a) [0x4873ca]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi(main+0x5f) [0x43f2af]
/lib64/libc.so.6(__libc_start_main+0xf5) [0x7fe4eae8f555]
/s/emib/s/relion/v3.1/beta6/bin/relion_refine_mpi() [0x441aa5]
==================
ERROR: 
helix.h::HelicalSegmentPriorInfoEntry::operator<(): A pair of same helical segments is found!

[biochem-fan]

Thanks for your report.

Can you try an old commit (git checkout 84d160f)? This might be related to a new helical prior introduced recently.

[acook2]

Thanks very much for your help.

We don't have that exact version on our system, but the error still occurs with the oldest v3.1 commit we have (b86482).