Scout: Scout user meeting agenda 9/9 -20

[henrikstranneheim]

Welcome to the Scout user meeting @9:00-10:30 am!

Location: Franzénrummet NKS J5-13 (Solna)

However, due to the current situation we will aim to host this on zoom.

Please help us fill the agenda with topics that may interest you.

Agenda

1. Demo of new features since last meeting
2. Feature requests
3. Pressing issues to prioritise (for reasons we are unaware of 😊)?
4. Upcoming Scout features
5. New MIP and Balsamic releases - what can we expect this time, and about when?

[dnil]

Reserved Franzénrummet NKS J5-13 (Solna). This room seats 20 persons in non-pandemic, and would currently be rated for maybe seven to ten. A few of us will go there, but we will try to stick mostly to Zoom. Link to follow!

[dnil]

Most welcome!

https://ki-se.zoom.us/j/64788459154

[heronikdin]

Hi, from cust 087 (hematology)!

I have some points I would like to bring up for discussion. Sorry in advance if this is not the forum for all them, and for a long message coming ahead!

1. For some samples there are many genes in the variant list. It is very hard to distinguish if a variant is true or not if there are so many genes within one column. I have discussed this with Hassan, and apparently there are some transcript errors causing this. Is there any way to come around this problem? (ex: "bossponge variant nr 1126).
2. We need to see see the VAF once the variant is pinned, because if not I have to go back to the complete variant list and then find my variant among the many pages. OR: if VAF could be seen in the GT call square as well, instead of the genotype quality -1?
3. There are many transcript errors, and the positions are then not right. For eg if a variant comes up in the variant list and then you click in on it, its not always the primary HGNC transcript which have been displayed in the variant list.
4. There is no “previous page”, only “next page” or first page”. It would be nice to have a "previous page" as we have so many variants to go through.
5. Every time I refresh the variant list page or go from a variant to the variant list, I need to reload my filter for variants. The back button does not work, brings me to an invalid page.
6. Several of the low frequency variants come up in several patients (I have a complete list of them all in an excelsheet if anyone is interested). For now I mark them as “common local” and choose a public comment. But can we somehow “hide” them in the future?

See you next week / Hero

[dnil]

Thank you @heronikdin! Excellent! To me these points seem mature enough to just make separate issues from? We can be your secretaries this time if you wish. But I just have to write a longer response.. 😜

1. Fairly sure this happens already before Scout loading, and likely in VEP so hard for us to control. But, we could perhaps try to highlight the genes that were on the panel and caused the variant to show?
   https://github.com/Clinical-Genomics/scout/issues/2068

2. Two issues here:

• 2a we can put that in the pinned box on the case page perhaps? And by "VAF" you actually mean the ratio of observed vs total alleles on that position, right? Anywhere else? I guess on the variant page it is somewhat obvious from the gt square, but do you want it written out there as well perhaps?
  https://github.com/Clinical-Genomics/scout/issues/2069

• 2b is for Balsamic interaction: GQ should not be -1, it should reflect variant call quality. I'm sure several similar measures are produced - lets get them imported. Presumably it is intended to cover this in https://github.com/Clinical-Genomics/scout/issues/1245?

1. There is on one hand not much we can do about genes with transcript errors. On the other, that happens exceedingly seldom for the RD genes. Any particular examples would be welcome and we can perhaps understand better what issue might be at play!

2. Sure, that should not be so difficult. We'll fix. https://github.com/Clinical-Genomics/scout/issues/2067
   It is however very seldom needed for RD for one very good reason: ranking. I strongly suggest you once again investigate that line of thought with e.g. Hassan.

3. Back works, sort of, for the browsers I use. I assume this is an IE or Edge issue? I generally encourage using the "open in other tab" function, keeping one "main filter" page open, then multiple tabs with one open on each variant on the list, to close when you have triaged the variant. That said, lets open an issue and talk about what we can do about propagating filter status. We have some more tools now than when we last looked at that.

4. So basically that is an issue for the underlying pipeline: scout only shows what it gets. If the local db used is good, you should not have this. That said, we developed loqus for that reason, and keeping a separate cancer-loqus will be possible from the next release of scout, thanks to @mikaell, @moonso and others.

[heronikdin]

Hi Daniel and thank you for the response!

I actually asked Henrik if I should create issues for all of them or if I should write everything in the agenda. The suggestion was to write everything up for now, and then we could maybe discuss them in the meeting next week. Which I actually agree on, so if you don’t mind, can we please discuss some of the points below and I can also find examples to clarify some of the problems more?

By the way, I do not mind creating issues at all 😊

[dnil]

Raw list of changes since last time - TL;DR

[4.21 - release candidate TBA]

Added

• Support to configure LoqusDB per institute
• Highlight causative variants in the variants list
• Add tests. Mostly regarding building internal datatypes.
• Remove leading and trailing whitespaces from panel_name and display_name when panel is created
• Mark MANE transcript in list of transcripts in "Transcript overview" on variant page
• Show default panel name in case sidebar

Fixed

• Report pages redirect to login instead of crashing when session expires
• Variants filter loading in cancer variants page
• User, Causative and Cases tables not scaling to full page
• Improved docs for an initial production setup
• Compatibility with latest version of Black
• Fixed tests for Click>7
• Clinical filter required an extra click to Filter to return variants
• Restore pagination and shrink badges in the variants page tables
• Removing a user from the command line now inactivates the case only if user is last assignee and case is active
• Bugfix, LoqusDB per institute feature crashed when institute id was empty string
• filter removal and upload for filters deleted from another page/other user

Changed

• Highlight color on normal STRs in the variants table from green to blue
• Display breakpoints coordinates in verification emails only for structural variants

[4.20 - undeployed]

Added

• Display number of filtered variants vs number of total variants in variants page
• Search case by HPO terms
• Dismiss variant column in the variants tables
• Black and pre-commit packages to dev requirements

Fixed

• Bug occurring when rerun is requested twice
• Peddy info fields in the demo config file
• Added load config safety check for multiple alignment files for one individual
• Formatting of cancer variants table

Changed

• Updated the documentation on how to create a new software release
• Genome build-aware cytobands coordinates
• Styling update of the Matchmaker card
• Select search type in case search form

[4.19 -- current live instance in CG Solna]

Added

• Show internal ID for case
• Add internal ID for downloaded CGH files
• Export dynamic HPO gene list from case page
• Remove users as case assignees when their account is deleted
• Keep variants filters panel expanded when filters have been used

Fixed

• Handle the ProxyFix ModuleNotFoundError when Werkzeug installed version is >1.0
• General report formatting issues whenever case and variant comments contain extremely long strings with no spaces

Changed

• Created an institute wrapper page that contains list of cases, causatives, SNVs & Indels, user list, shared data and institute settings
• Display case name instead of case ID on clinVar submissions
• Changed icon of sample update in clinVar submissions

[4.18]

Added

• Filter cancer variants on cytoband coordinates
• Show dismiss reasons in a badge with hover for clinical variants
• Show an ellipsis if 10 cases or more to display with loqusdb matches
• A new blog post for version 4.17
• Tooltip to better describe Tumor and Normal columns in cancer variants
• Filter cancer SNVs and SVs by chromosome coordinates
• Default export of Assertion method citation to clinVar variants submission file
• Button to export up to 500 cancer variants, filtered or not
• Rename samples of a clinVar submission file

Fixed

• Apply default gene panel on return to cancer variantS from variant view
• Revert to certificate checking when asking for Chanjo reports
• scout download everything command failing while downloading HPO terms

Changed

• Turn tumor and normal allelic fraction to decimal numbers in tumor variants page
• Moved clinVar submissions code to the institutes blueprints
• Changed name of clinVar export files to FILENAME.Variant.csv and FILENAME.CaseData.csv
• Switched Google login libraries from Flask-OAuthlib to Authlib

[4.17.1]

Fixed

• Load cytobands for cases with chromosome build not "37" or "38"

[4.17]

Added

• COSMIC badge shown in cancer variants
• Default gene-panel in non-cancer structural view in url
• Filter SNVs and SVs by cytoband coordinates
• Filter cancer SNV variants by alt allele frequency in tumor
• Correct genome build in UCSC link from structural variant page

Fixed

• Bug in clinVar form when variant has no gene
• Bug when sharing cases with the same institute twice
• Page crashing when removing causative variant tag
• Do not default to GATK caller when no caller info is provided for cancer SNVs

[4.16.1]

Fixed

• Fix the fix for handling of delivery reports for rerun cases

[4.16]

Added

• Adds possibility to add "lims_id" to cases. Currently only stored in database, not shown anywhere
• Adds verification comment box to SVs (previously only available for small variants)
• Scrollable pedigree panel

Fixed

• Error caused by changes in WTForm (new release 2.3.x)
• Bug in OMIM case page form, causing the page to crash when a string was provided instead of a numerical OMIM id
• Fix Alamut link to work properly on hg38
• Better handling of delivery reports for rerun cases
• Small CodeFactor style issues: matchmaker results counting, a couple of incomplete tests and safer external xml
• Fix an issue with Phenomizer introduced by CodeFactor style changes

Changed

• Updated the version of igv.js to 2.5.4

[4.15.1]

Added

• Display gene names in ClinVar submissions page
• Links to Varsome in variant transcripts table

Fixed

• Small fixes to ClinVar submission form
• Gene panel page crash when old panel has no maintainers

[4.15]

Added

• Clinvar CNVs IGV track
• Gene panels can have maintainers
• Keep variant actions (dismissed, manual rank, mosaic, acmg, comments) upon variant re-upload
• Keep variant actions also on full case re-upload

Fixed

• Fix the link to Ensembl for SV variants when genome build 38.
• Arrange information in columns on variant page
• Fix so that new cosmic identifier (COSV) is also acceptable #1304
• Fixed COSMIC tag in INFO (outside of CSQ) to be parses as well with & splitter.
• COSMIC stub URL changed to https://cancer.sanger.ac.uk/cosmic/search?q= instead.
• Updated to a version of IGV where bigBed tracks are visualized correctly
• Clinvar submission files are named according to the content (variant_data and case_data)
• Always show causatives from other cases in case overview
• Correct disease associations for gene symbol aliases that exist as separate genes
• Re-add "custom annotations" for SV variants
• The override ClinVar P/LP add-in in the Clinical Filter failed for new CSQ strings

Changed

• Runs all CI checks in github actions

[4.14.1]

Fixed

• Error when variant found in loqusdb is not loaded for other case

[4.14]

Added

• Use github actions to run tests
• Adds CLI command to update individual alignments path
• Update HPO terms using downloaded definitions files
• Option to use alternative flask config when running scout serve
• Requirement to use loqusdb >= 2.5 if integrated

Fixed

• Do not display Pedigree panel in cancer view
• Do not rely on internet connection and services available when running CI tests
• Variant loading assumes GATK if no caller set given and GATK filter status is seen in FILTER
• Pass genome build param all the way in order to get the right gene mappings for cases with build 38
• Parse correctly variants with zero frequency values
• Continue even if there are problems to create a region vcf
• STR and cancer variant navigation back to variants pages could fail

Changed

• Improved code that sends requests to the external APIs
• Updates ranges for user ranks to fit todays usage
• Run coveralls on github actions instead of travis
• Run pip checks on github actions instead of coveralls
• For hg38 cases, change gnomAD link to point to version 3.0 (which is hg38 based)
• Show pinned or causative STR variants a bit more human readable

[4.13.1]

Added

Fixed

• Typo that caused not all clinvar conflicting interpretations to be loaded no matter what
• Parse and retrieve clinvar annotations from VEP-annotated (VEP 97+) CSQ VCF field
• Variant clinvar significance shown as not provided whenever is Uncertain significance
• Phenomizer query crashing when case has no HPO terms assigned
• Fixed a bug affecting All SNVs and INDELs page when variants don't have canonical transcript
• Add gene name or id in cancer variant view

Changed

• Cancer Variant view changed "Variant:Transcript:Exon:HGVS" to "Gene:Transcript:Exon:HGVS"

[4.13]

Added

• ClinVar SNVs track in IGV
• Add SMA view with SMN Copy Number data
• Easier to assign OMIM diagnoses from case page
• OMIM terms and specific OMIM term page

Fixed

• Bug when adding a new gene to a panel
• Restored missing recent delivery reports
• Fixed style and links to other reports in case side panel
• Deleting cases using display_name and institute not deleting its variants
• Fixed bug that caused coordinates filter to override other filters
• Fixed a problem with finding some INS in loqusdb
• Layout on SV page when local observations without cases are present
• Make scout compatible with the new HPO definition files from http://compbio.charite.de/jenkins/
• General report visualization error when SNVs display names are very long

[ViktorHy]

Hi! Is this a general meeting or just for Stockholm? I know there's been talk about a user meeting also in Lund around october.

[dnil]

All users are most welcome! While the pandemic got in the way of this springs intended all-hands meetup/workshop, this is a nice opportunity to get input from all sites!

[dnil]

4.21 walkthrough highlights:

Case

• Show default panel name in case sidebar
• Warn users if case default panels are outdated (clean the browser cache if popover looks weird)
• Show somatic VAF for pinned and causative variants on case page
• User, Causative and Cases tables not scaling to full page
• Select search type in case search form
• Show internal ID for case
• Add internal ID for downloaded CGH files
• Export dynamic HPO gene list from case page
• Keep variant actions (dismissed, manual rank, mosaic, acmg, comments) upon variant re-upload
• Keep variant actions also on full case re-upload
• Always show causatives from other cases in case overview

Variants

• Highlight causative variants in the variants list
• Use institute-specific gene panels in variants filtering - always use latest gene panel in filtering - but case may be analysed with older version
• Keep variants filters panel expanded when filters have been used
• Display number of filtered variants vs number of total variants in variants page
• Previous buttons for variants pagination
• Formatting of cancer variants table
• Filter SNVs, SVs and cancer variants by cytoband coordinates
• Filter cancer SNVs and SVs by chromosome coordinates
• Filter cancer SNV variants by alt allele frequency in tumor
• Tooltip to better describe Tumor and Normal columns in cancer variants
• Button to export up to 500 cancer variants, filtered or not
• COSMIC badge shown in cancer variants
• Adds verification comment box to SVs (previously only available for small variants)
• Cancer Variant view changed "Variant:Transcript:Exon:HGVS" to "Gene:Transcript:Exon:HGVS"
• Highlight color on normal STRs in the variants table from green to blue (might require cleaning of the browser cache)

ClinVar

• Display case name instead of case ID on clinVar submissions
• Rename samples of a ClinVar submission file
• Display gene names in ClinVar submissions page
• Changed name of clinVar export files to FILENAME.Variant.csv and FILENAME.CaseData.csv

Variant

• Show an ellipsis if 10 cases or more to display with loqusdb matches
• ClinVar SNVs track in IGV
• Clinvar CNVs IGV track (discussion point..)
• Links to Varsome in variant transcripts table
• Variant loading assumes GATK if no caller set given and GATK filter status is seen in FILTER

Institute

• Define institute-specific gene panels for filtering in institute settings
• Created an institute wrapper page that contains list of cases, causatives, SNVs & Indels, user list, shared data and institute settings

Panels

• Gene panels can have maintainers

CLI, user facing implications

• Remove users as case assignees when their account is deleted
• Removing a user from the command line now inactivates the case only if user is last assignee and case is active

Many (many..) bugs fixed

• several missing hg38-adaptions (thank you Lund!)
• better handling of delivery reports for rerun cases
• a few ClinVar filter/loading issues; some issues with old references remain. MIP9!
• Fixed a problem with finding some INS in loqusdb
• very long comments rendered in reports

Upcoming

• Add SMA view with SMN Copy Number data - pending MIP9
• Chromograph - pending HK update and possibly MIP update
• Managed variants - next release?
• Genepanels extending into STRs - Pending update to Stranger
• Custom IGV tracks
• Easier deployment via podman / systemd - pilot with course?

[hassanfa]

BALSAMIC changes since previous user meeting that affect final clinicians:

• New coverage tool: mosdepth (replaced Sambamba)
• CRAM files for TGA and WES runs
• Annotated germline calls
• CNVs are output as VCF (beside PDF)
• Multiqc report in Scout
• Draft for pre-Scout filtering
• Lots of optimization, software updates, and codebase improvements

Roadmap:
From previous user meeting that are not yet implemented:

• Additional annotation sources (e.g. TCGA)
• Show CNV results in Scout (ready, needs Scout integration)
• MSI results in Scout (ready, needs in house data flow integration)

[henrikstranneheim]

MIP

Version 9.0 (currently in validation)

• Updated references (clinvar. loqus (>3000 WGS))
• New splicing annotation using spliceAI
• Smncopynumbercalling per family
• Call CYP2D6 alleles with star_caller from the Cyrius package
• Added telomerecat analysis for estimating telomere length from wgs
• Support bwa_mem2 for faster alignment with identical results
• Many software updates (including GATK)
• New workflow for processing small panels (not implemented in production at Clinical Genomics yet)

Roadmap

• Last release on genome build grch37
• Get RNA-seq into production at Clinical Genomics