general impressions of v3

[maglott]

predicted protein descriptions: compare the predicted protein descriptions for https://v2.mutalyzer.nl/name-checker?description=%09NM_001204398.1%3Ac.268_288dup and https://mutalyzer.nl/normalizer/%09NM_001204398.1:c.268_288dup.
I find the former much easier to use to evaluate what has changed.

response time The response time for v2 is much better for simple normalizations. Is it possible to wait to compute the data necessary to support the new features until they are requested?

variant sequence overview My personal preference is the compact display in v2. I prefer to scroll as little as possible.

That said, I appreciate the new features

[erikzmuda]

Came here to provide feedback on the v3 UI. Strongly dislike the update. The query response should show the info you're after without having to expand all of the sub tabs. It is way less efficient to use now. Really wish there was a way to control the version being used b/c v3 is a huge step in the wrong direction from user perspective.

[jfjlaros]

The response time for v2 is much better for simple normalizations.

This has two main reasons.

1. Mutalyzer 3 uses the Description Extractor to normalise variant descriptions. This allows for the normalisation of allele descriptions, e.g., NG_012337.3(NM_003002.4):c.1_10delinsTGGCGGTT is normalised to NG_012337.3(NM_003002.4):c.[1del;10del]. We can now guarantee that whenever two variant descriptions result in the same sequence (after applying them to the same reference sequence), they will result in the same normalised description. Unfortunately, this added correctness comes with a performance penalty.
2. Mutalyzer 3 currently has a very rudimentary caching system. This is being worked on.

We will take your comments about the visualisations into account in our discussions about the interface.

[jfjlaros]

Really wish there was a way to control the version being used

Mutalyzer 2 is still online, but perhaps you can help us improve the new interface instead.

For example, would you be helped if we added a configuration option that allows the user to store some preferences about which folds should be open by default?

Your suggestions are welcome.

[maglott]

I would welcome options to control (1) what is shown (open) by default and (2) the order of presentation. I am accustomed to checking not only the HGVS expression but the display of the reference/alternate sequence that expression represents, and so am still using v2 for simple deletions, insertions, and SNV. In v2 these appear without clicks and scrolling.
I personally use NCBI's graphic sequence displays to view the location of variants across all RefSeqs and am less likely to use the tool to display those mappings.

[erikzmuda]

Hi! Thank you so much for being open to feedback. Our group really values and relies on the mutalyzer service. For context, I am one of the lab directors working within the clinical genomics group at Myriad Women's Health. We perform carrier screening on healthy adults for the purpose of informing couples of their reproductive risk across a panel of ~200 autosomal recessive diseases. Our group is made up of ~20 curators working full time to assess the impact of the variants we detect in this population. Elizabeth Collins (copied on this reply) and I have collected feedback on v3 from individuals within our group and from former employees now working elsewhere. Overall, there were two main points we wanted to comment on. 1. Position Converter.... We've found that the input has become very difficult to use; almost unusable. The data entry fields are confusing and would benefit from some helper text to clearly define expected value and if required. Unfortunately, we're having so much trouble with the current format that we are not able to use it at all anymore. While definition hover-overs with examples for each field might help a little, we would really prefer a single field where you can enter your query (we're often copy/pasting from a different source as single string). The query is now split up across 6 fields, with the last two fields asking for the unknown output we're trying to obtain through conversion. Here's an example that gave us exactly what we needed in v2 (g. converted to c. on all possible transcript).... We are unable to run this query on v3. 2. [cid:6ed4a352-03af-41d1-aa0f-b99f3939db1d] 3. Name Checker (now called Normalizer).... If I'm being honest with you, we had a hard time finding a way to craft constructive feedback for this page. A full reversion to v2 layout/functionality is highly desired. The way the v2 page is designed was near perfect in our opinion. The effective use of space, the layout of critical elements we're interested in, the speed of loading... everything. It was perfect. v3 is much less efficient to use (based on having to expand tiles and rows for exons) and it appears to be locked to a genome build that we aren't yet using which is known to impact some of our transcript versions. The page aesthetics and use of hg38 will ultimately make it difficult for us to continue using the v3 tool. Thanks again for being open to feedback. I know it is hard to create a service/tool that meets everyone's needs, and we really do appreciate the service you've provided. Warmest Regards, Erik Zmuda, PhD, FACMG, CGMB Laboratory Director | Myriad Women's Health 180 Kimball Way, South San Francisco, CA 94080 Remote: Columbus, Oh ***@***.*** | www.myriad.com<http://www.myriad.com/> [cid:d20003ff-aa44-4d60-a5b2-319e3a584558]

…

________________________________ From: maglott ***@***.***> Sent: Friday, October 28, 2022 6:03 PM To: mutalyzer/website ***@***.***> Cc: Erik Zmuda ***@***.***>; Comment ***@***.***> Subject: Re: [mutalyzer/website] general impressions of v3 (Issue #15) I would welcome options to control (1) what is shown (open) by default and (2) the order of presentation. I am accustomed to checking not only the HGVS expression but the display of the reference/alternate sequence that expression represents, and so am still using v2 for simple deletions, insertions, and SNV. In v2 these appear without clicks and scrolling. I personally use NCBI's graphic sequence displays to view the location of variants across all RefSeqs and am less likely to use the tool to display those mappings. — Reply to this email directly, view it on GitHub<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fmutalyzer%2Fwebsite%2Fissues%2F15%23issuecomment-1295511632&data=05%7C01%7Cerik.zmuda%40myriad.com%7C28f259b7093743a6730f08dab930337d%7Cf2d22b33b2fe4c0c8f5f221652c60157%7C0%7C0%7C638025913924025475%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=h9pHisREt5p4hPRWj0GTY6rgpcQO2ndJJwaRPbYN4UA%3D&reserved=0>, or unsubscribe<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAQ6WIKSNA2E7GTKKTLPHJZTWFRER3ANCNFSM6AAAAAARRB3SL4&data=05%7C01%7Cerik.zmuda%40myriad.com%7C28f259b7093743a6730f08dab930337d%7Cf2d22b33b2fe4c0c8f5f221652c60157%7C0%7C0%7C638025913924025475%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=9ovgMuy46gfmoYNDyVxh5KywqNcUh7QIn5gyq%2Bl1Qxw%3D&reserved=0>. You are receiving this because you commented.Message ID: ***@***.***>

[jfjlaros]

Thank you and your group for the detailed feedback. Please allow me to give some background information and motivation before proceeding to a discussion about possible improvements.

---

The interface of the Position Converter was changed intentionally because the former interface gives people the impression that HGVS descriptions are mapped or lifted over from one reference sequence to an other, while in reality only the positions were converted. This resulted in:

• Invalid descriptions being converted to other invalid descriptions.
• Correct descriptions being converted to wrong or invalid descriptions (notably conversions to and from transcripts residing on the negative strand were problematic).

We have always recommended to use the Name Checker on the output of the Position Converter, but this advice was rarely followed. For more information and examples, please see issue #14.

In Mutalyzer 3 we aim to have the functionality of both the Mutalyzer 2 Name Checker as well as the Mutalyzer 2 Position Converter integrated in the Normalizer.

---

From your feedback I think we can extract three technical questions:

1. Can we provide a workflow that is similar to the Mutalyzer 2 Position Converter workflow, but without all of its flaws?
2. Can we improve the responsiveness of Mutalyzer 3?
3. Can we improve the layout of the Normalizer results page?

To address point 1, we could think about allowing a combination of a genome build and a RefSeq identifier as input for the Normalizer. For example:

GRCh37(NM_002001.4):c.10del

Which would be corrected to:

NC_000001.10(NM_002001.4):c.10del

The "converted" descriptions will be available in the "Equivalent Descriptions" fold of the results page.

Point 2 is being worked on as we speak.

To address point 3, I would suggest to open a separate discussion to get more feedback from the community and to share ideas for improvement.

[maglott]

Thanks for this detailed response. For point 3, was there a summary of use cases used in the design?
My most frequent usages (I am currently going through old papers for OMIM) are to enter an HGVS expression

• and look at what is now view variants sequence overview to determine if the effect of the change is captured correctly by the HGVS
• and check for a warning that the variant includes splice sites
• verify changes in restriction sites

[jfjlaros]

@erikzmuda

Point 1 has been addressed. You may want to try it here.