<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body style="overflow-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;">We recently had a case where something like this happened because some movies had fewer frames than others (for reasons that we do not understand, but that does not really matter here). Setting the max. frame number in polishing in a way that all movies satisfied that criterion solved the issue. Maybe worth checking.<div><br></div><div>Basil</div><div><br id="lineBreakAtBeginningOfMessage"><div><br><blockquote type="cite"><div>Am 27.04.2026 um 23:50 schrieb Erik Hartwick via 3dem <3dem@ncmir.ucsd.edu>:</div><br class="Apple-interchange-newline"><div>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<div><p class="p1" style="direction: ltr; text-align: left; text-indent: 0px; text-transform: none;">
<span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">Hello 3DEM community,</span></p><p class="p2" style="text-align: left; text-indent: 0px;"><span style="font-size: 16px;">I am reaching out to see whether anyone has encountered a similar issue with Bayesian polishing in RELION 5 via SBGrid. I have run polishing on the same
</span>“<span style="font-size: 16px;">system” for another project without issue; however, this project seems to be problematic. </span></p><p class="p2" style="text-align: left; text-indent: 0px;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt; text-transform: none;">We are processing an SPA dataset in RELION 5.0.1. Refinement behaves normally,
but during Bayesian polishing a very large fraction of particles is consistently removed (~50%). The job completes far enough to report the issue, but polishing repeatedly rejects roughly half of the particles. This was a large collection 17k images at 130kx
<s>(</s>~0.9apix), 300kv, 100A target dimension, </span>preferred orientation is strong, collection was performed with and without tilt, and lastly Blush is being used to help with the orientation bias. </p><p class="p2" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">What we are seeing:</span></p>
<ul style="text-align: left;">
<li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">upstream particle set refines normally at 1xbin box 256 pix</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">Bayesian polishing then removes about 133k particles from a starting set of about 261k</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">this has happened repeatedly across multiple attempts</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">one representative message was:</span></p><p class="p2" role="presentation"><span style="text-transform: none;">WARNING: 136430 particles in the optics group 1 were removed</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">we also saw:</span></p><p class="p2" role="presentation"><span style="text-transform: none;">DEBUG: FC_WEIGHT didn't match</span></p><p class="p1" role="presentation"><span style="text-transform: none;">although this same debug line also appeared in MotionCorr stderr, so I am no longer sure it is directly relevant</span></p>
</li></ul><p class="p2" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">What we have already tried:</span></p>
<ul style="text-align: left;">
<li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">reran MotionCorr from scratch</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">reran CTF estimation</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">reran micrograph selection</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">re-extracted particles</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">reran refinement</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">traced the lineage backward through the pipeline to look for an earlier metadata issue</span></p>
</li></ul><p class="p2" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">What we found:</span></p>
<ul style="text-align: left;">
<li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">the problem does not appear to be explained by missing micrograph-path matches between the particle STAR and corrected_micrographs.star</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">refinement continues to work on these particle sets</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">the issue seems to become visible only at polishing</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">the problematic lineage appears to emerge after a 3D classification / selection branch, where a parent stack of about ~1M particles was narrowed to a subset of about ~250k particles</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">downstream versions of that selected subset continue to show the polishing rejection problem</span></p>
</li></ul><p class="p2" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">A few additional notes:</span></p>
<ul style="text-align: left;">
<li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">this is not normal behavior in our hands, and I have not encountered it before with this workflow</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">MotionCorr does report some movies where RELION could not fit a reasonable local motion model and used global trajectory only, but I am not sure whether that is related</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">we also saw repeated warnings during extraction about some micrographs having very few or no particles after selection, but regrouping did not appear to explain the polishing rejection</span></p>
</li></ul><p class="p2" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">My main questions are:</span></p>
<ol start="1" style="text-align: left;">
<li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">Has anyone seen Bayesian polishing repeatedly remove such a large fraction of particles while refinement still works normally?</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">Are there known metadata inconsistencies or STAR-file issues that can survive refinement but cause polishing to reject particles?</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">Has anyone seen this associated with a specific 3D classification / selection lineage?</span></p>
</li><li style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;"><p class="p1" role="presentation"><span style="text-transform: none;">Are there RELION 5-specific polishing issues with re-extraction, optics groups, or movie metadata that we should inspect more carefully?</span></p>
</li></ol><p class="p1" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">Any suggestions on where to look next would be greatly appreciated. Efforts
to start with a new processing attempt have started but knowing why this occurred would be great. </span></p><p class="p1" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">All the best,</span></p><p class="p1" style="text-align: left; text-indent: 0px; text-transform: none;"><span style="font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">Erik</span></p>
<div style="direction: ltr; font-family: Aptos, Arial, Helvetica, sans-serif; font-size: 12pt;">
<br>
</div>
<div id="ms-outlook-mobile-signature">
<div class="MsoNormal" style="font-family: Arial, sans-serif;">Erik Hartwick PhD</div>
<div class="MsoNormal" style="font-family: Arial, sans-serif;">Director BioKEM Core Research Facility</div>
<div class="MsoNormal" style="font-family: Arial, sans-serif;">Department of Biochemistry</div>
<div class="MsoNormal" style="font-family: Arial, sans-serif;">University of Colorado Boulder</div>
<div class="MsoNormal" style="font-family: Arial, sans-serif;"> </div>
<div class="MsoNormal" style="color: rgb(191, 191, 191);"> </div>
</div>
</div>
_______________________________________________<br>3dem mailing list<br>3dem@ncmir.ucsd.edu<br><a href="https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem">https://mail.ncmir.ucsd.edu/mailman/listinfo/3dem</a><br></div></blockquote></div><br></div></body></html>