• [X] All headers should be left justified. Please avoid centering headers of any type. Why: Because all windows are made smaller from the right to the left. Centered headers therefore "squirrel" about depending on window width. We are not designing wedding invitations.

• [X] Do not use ":" in general do symbol after headers [Unclear]

• [X] Careful please with the left margin. There must always be some white space at the far left.

• [ ] New request: Could we have a new field called "*Curator Notes*" We have been using the "*Abstract*" field for curation notes but this is unwise since the Abstract will erase the notes. Here is an example of how Suheeta and I use the Abstract field inappropriately.

• [X] It would be great to "populate" the fields with sample text (greyed out perhaps) so that users know how to enter the Authors and other fields. Authors should be entered as shown here (NCBI format: Roy S, Ingels J, Bohl CJ, McCarty M, Lu L, Mulligan MK, Mozhui K, Centeno A, Williams EG, Auwerx J, Williams RW), but nine times out of ten new users use their own format.

• [ ] Along the lines of item 5 -- each field should have a "help" button to provide guidance on format. For example, how does one enter *Units*? What the heck is the difference between *Pre Publication* and *Post Publication* *Descriptions* and *Abbreviations? *What is the difference between *Submitter* and *Owner *and* Authorized Users* in terms of priviledges? How many *Owners* can there be? Does the submitter name have to GN-recognized email or is *sroy12* good enough?

• [ ] If the user enters 100 traits for a study, are those 100 traits all linked in some way that they can easily be edited jointly? GN1 did this and if the user updated the *PubMed ID* for one record then all records with the same PubMed ID received the matched data (for example *Journal*, *Pages*, *Month*, *Year*, *Title*, *Abstract*. Now this is tricky and could require the user to manually enter the *PubMed ID* 100 times and we want to avoid that. For this reason we need to add a new field on the *Experiment ID* so that all 100 traits are assigned the same *Experiment ID *upon original entry and long before there is a *PubMed ID.*

• [X] Note the "m" in PubMed should be capitalized. And ideally there would be a hot link for the *PubMed ID* in our *Edit Trait for Published Database*

• [X] Hmm, this header is not very good-- *Edit Trait for Published Database.* a. We mean the word "edit" as a verb, but status of this word is ambiguous. b. Published Database is also not right, since the intent is to allow the user/owner to control access. Maybe this header would be better-- *Trait Metadata and Data Editing Form*

• [X] The *Submit* and *Reset* buttons are the same color and much to close together. Easy to click on the wrong button and waste your time. Please use green and red colors to make this less equivocal. And put the buttons on the page at both the top and at the bottom as in GN1 to save the user time.

• [ ] The section of the page in which the user uploads the vector of data is a case study of extreme subtlety and also a mess in Safari. The data entry is obviously critical and should be available at the TOP and BOTTOM of the page. And note some other strange formatting in the footer.

- "Curator Notes" and "Abstract"; we'd have to add a new column for this. - For the help button, I'll reach out to Zach and see if there's a "description" field somewhere we can use. - Working with Safari is challenging... Also, for the upload button, I'll check that later... What happens if someone uses both buttons? I haven't investigated that.

Fixes [so far]:

• Metadata Upload UI Improvements

• Style improvements for editing phenotypes

Use-case (mRNA assay):

PURPOSE: I want to update and correct the gene symbol and other metadata for ALL probesets that are associated with the same geneID - 170791 in this case. This mouse gene is now known as Rbm39. It used to be called Rnpc2. This is now wrong and needs to be fixed across all probesets associated with the geneID 170791.

Step 1. I search the default database (BXD hippocampus today) using the search term Rbm39 and I find 14 records within this single probesetfreeze. We have many many probesetfreezes and many will also have the incorrect gene symbol which I now want to correct.

Obviously I can not fix them all one-by-one. GN1 solved this problem many years ago by running a script and SQL update that used the GeneID to fix all fields that were updated. GN2 code is not aware of this challenge and that makes the GN2 curation not useful to me or other curators for this crucial function.

Step 2. I open the top Record ID 1420982_at above.

Step 3. I edit some of the fields. In this case I have correct the symbol to read Rbm39. I also modified the aliases list and I also added the homologene ID. This information needs to be updated jointly in every probeset that has the identical GeneID number. That will make a global correction possible. Have a look at how GN1 handles this.

Step 4. Now the user is ready to submit their corrections. To do so the poor user must scroll to the very bottom of the page. That is not good UX. Another problem we have discussed several times - the two buttons with different function - Submit Change and Reset are next to each other. If the user/curator is working quickly there is a risk that they hit the Reset function. Do we need a Reset function? I don't think so.