-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
VEP deployment - multiple versions on same disk #1057
Comments
I updated the VEP wiki to explicitly have versions in the VEP paths Also modifying pipeline to pass cache and plugins dir so we can have diff versions of plugins on the same machine
|
Installed 110 in the right place on vg test, vg prod and Shariant shared drive. Altered settings so it should work and use this next upgrade Check whether it works via looking at pipeline runs |
I've updated shariant prod setting to be VEP 108 |
tested - run annotations etc on vg test and works fine |
* Panal App VUS/GUS relationship download * Panal App VUS/GUS relationship download * Providing methods to find more useful relationships between term and gene symbol for report * issue #2647 - liftover refactor * issue #2647 - start of bcftools liftover * Issue #2647 - BCFTools liftover * Issue #2647 - Remove vestigial NCBI remap traces * Issue #2647 - enable in vg test * Stop warning that doesn't apply here * Issue #2647 - be able to create variant for long ones... * This occasionally failed - execution continued after asking to change window * issue #1054 - make bcftools liftover part of standard ANNOTATION settings * issue #1054 - misc liftover issues * Issue #1052 - analysis template version * Make VEP have version in path * Shariant test config - enable BCFTools liftover * Upload initially shows error wrongly in pipeline race condition * issue #818 - don't uniq on preprocess anymore * issue #1056 - vcf_clean_and_filter convert contig headers * issue #1057 - VEP deployment change to explicit version * vg test config * issue #980 - karyomapping use symbol instead of gene * Renamed counts to Counts(germline only) on overlaps page (#1046) * Fix issue where alleleOriginToggle was being called with undefined * WIP fixing conda dependencies * Update of conda environment (works again in conda) * Add new evidence key for somatic testing * Show better quick clinical significance values (for somatic and germline) * Fix recently introduced bug in suggested terms for an allele * Allow link for condition matching to appear on a classification even when not in edit mode * issue #1059 - all variants node * issue #1059 - all variants node * Adjust node counts a little * issue #1043 - export VCF * Tidy up server status page removing redundant data (now go to Overall Status) * Fix the evidence key values * Get c.HGVS showing properly on variant details page again * issue #758 - configure quick links via settings * More comments in conda file * Move condition text match from classification.html to JSON to keep things in sync More requirements on when you will be linked to condition resolution * Style fixes for new liftover * Put liftover date on form * Move overall data to the splash server status page * Add Clinical Trails gov to quick links * Add liftover to settings menu if variants menu disabled * Update shariant prod settings to use VEP v108 * Make liftover page side menu change based on settings * Better wording for changing clinical contexts within a discordance * Tighten up the handling of embedded card or modal (used for triage but affected other modals by mistake) * Better safety around condition text relationships * For view metrics use heading of "Users" not "User" * Add overall stats to liftover pages * Slight formatting on liftover note * Rework the internals of the view user activity reports * More styling on liftover runs page * More styling on liftover pages * Tiny column alignment issue on liftover * linting * Fix exclusion of blank searches in metrics * Subtle change of header wording for activity report * Improve the formatting of missing IDs when making a batch for ClinVar * Proper int to string handling for missing IDs * issue #758 - quick links via settings * Reworking of Lab Differences using new zippable functionality for ExportRow * Put the allele URL into the lab compare (I think it was there before) * Enable SEARCH_HGVS_GENE_SYMBOL_USE_MANE for shariant test * Also try enabling SEARCH_HGVS_GENE_SYMBOL * Fix bug where condition URL was calculated before seeing if we had condition text * Enable extra variant annotations for testing * Attempt at parsing OncogenicityClassification for pulling in ClinVar records * issue #1193 - analysis audit log * issue #1193 - analysis audit log * issue #1193 - analysis audit log * issue #1193 - analysis audit log * issue #1193 - analysis audit log * variantgrid_private#1193 - Audit log for analysis * Get audit log name right in requirements * issue #1060 - variant details page, create allele for all variants not just shorty clingen ones * Sometimes can't reload analysis #1061 * issue #1193 - disable audit log for template copies etc * issue #1193 - Audit log for analysis - click to expand and see JSON * issue #1193 - Don't audit node cloning (was causing tests to fail) * temp fix for #1053 - will work out data fix later * Better categorising of ClinVar's Somatic/Germline records * issue #1053 - cohort genotype versions - make common cohort version match cohort.version * linting - format whitespace * linting - remove unused imports * Unused file * linting - f string w/o interpolation * linting - use generator * Increase the ClinVar parser version to purge old cache * Add django-audit-log to conda environment * ClinVar REcords More work on parsing condition * issue #1053 - missing import, make sure we retrieve only 1 cgc per cohort * issue #877 - Export column - c_hgvs_compat * Add MeSH as a non-local ontology set * Ability to render multiple conditions against a single ClinVar record * Don't abbreviate ref/alt in VCF * Fix bug with filtering out non-human clinvar records * Count homo sapiens, homo-sapiens, homosapiens the same as human for ClinVar records * A script to close manually raised flags * Raise classification change flag for both Classification and Somatic Clinical Significance * Filter based on creator of flag * Added prefixed clinvar batch export CB_ and clinvar export CE_ search functionality (#1066) * issue #3604 - COSMIC search * Update changelog * Tidy up code for close flags, allow it to re-open flags * Bug fixes for the condition checking code * panalapp work * PanelApp Compare: Fixes to stop infinite recurssion, correct method signatures. * #3591 panel app export --------- Co-authored-by: TheMadBug <[email protected]> Co-authored-by: Dave Lawrence <[email protected]>
@davmlaw How should we test this? |
This is a deployment issue, so not really user testable Shariant test has been deployed with this, and annotations work since then I changed the description from "multiple versions on same machine" to "multiple versions on same disk" - to reflect what it really is, and why it was a issue for Shariant |
I've reviewed the layout for this - it's basically letting different versions of VEP data co-exist on the same disk (to stop Shariant Prod from breaking when Shariant Test next upgrades its version of VEP). Which is to say, I believe it's effectively already been tested. If after the deploy if Shariant Prod's annotations don't start working, we can raise a new issue. |
VG VEP install instructions assumed there would only be 1 VEP installed per system. This caused 2 issues:
Running wrong version on diff deployments
If you modify the VEP dir on 1 machine, it breaks another system that relied on it being a certain version (as eg code was not setup to handle that version of VEP). This caused prod issue https://github.com/SACGF/variantgrid_shariant/issues/158
Plugin breaking
We only passed "--dir" which assumes the plugins are installed in
${VEP_CACHE_DIR}/Plugins
- if plugins change over time - this causes errors egThe text was updated successfully, but these errors were encountered: