Feedbacks on the peptide search page

[ruiwanguk]

@jadianes @noedelta Very nice work to get the search working again! Below are some additional comments:

• Peptide sequences should be in a different colour instead of white
• Change the colour that represents the results
• I feel that the way we lay out the numbers can be improved, may be we can make them consistent with spectrum libraries?
• Change “Spectra” to “#Spectra"
• Merge "Prec. M/Z" and “Prec. Charge” into “ Precursor, and separate the values using “/"
• Precursor charge should be integer, this might need to be changed in the index as well
• Add “#Projects” to show the number of projects
• View as List is linking to the home page at the moment
• Breadcrumb needs to be changed to “PRIDE > PRIDE Cluster > Peptide Search"
• Pagination can be improved
• The three peptide sequence examples under the search box can not find any results
• It would be really nice to show number of species and number of modifications as well
• It would be cool if we could include a table view for the results, so people can easily sort the results
• It would be really nice to include sorting and filtering functionalities

[jgriss]

@jadianes @noedelta again, this is a 1000x improvement compared to the original page! Thanks!!!

some comments:

• I agree with Rui, white text on grey background is hard to read on my screen
• Use background colour to either visualize size or reliability
• I'd put the table view on top of my wish list
• merge precursor m/z + charge with the sequence in the title (f.e. ABCDE (@120 m/z, 1+)) - open to discussion
• otherwise I agree with Rui's list completely
• example sequences next to search box (or rather below) do not get any results

[jadianes]

@jgriss @ruiwanguk @noedelta I agree with most of the feedback.

About the colours, I removed them cause before they represented cluster quality. Since now we are showing just high quality clusters they don't make sense anymore. However I'm keen on using colour to code any other information.

Sorting and filtering controls are definitively needed. I hope somebody will take care of it at some point so they are ready for the millestone. Same with a table view if you really want it. About this, I'm not really sure is the best way to present search results. Tables are good for comparing numbers and doing calculations. They can also be sorted, true, but also can be lists. Even the grid that we have now can use relative sorting, although I agree that if we really see our search users doing strict sorting of results (something that most search engines don't have btw) something like a list is needed. But we need to think about the use case we are covering here:

• Are we implementing a search scenario where web and the user are involved in a dialog to find a particular cluster? In this case the user is interested in finding one or few results and then go to the Cluster details page. The search result page should clearly display a few very important results at the very top with a minimum of sorting, filtering, etc. They also need to really pop out. Tables are difficult or at least not "fun" to scan through. Since most of the data cells have the same size, etc. They look too uniform. They are very rigid in style and layout, lacking of visual hierarchy. It is not easy to drive the user to the right piece of information (e.g. showing the sequences or modifications more prominently). This is the typical use case of finding a particular product in a big collection (e.g. eBay, Amazon, or even Google).
• Are we providing the user with a set of results for him to compare or summarise? Then a table is very good. But then here there are other visual representations that are better for comparison (although tables are very good as I say).

In any case having different representations for the search results is a plus. I think that this could be a strong point for our site. But my opinion is that we should start showing a list of results (if we are not happy with the grid) and give the user the option of checking them also as a table or chart if she or he is interested in comparing results or knowing the context of the search results (e.g. a bubble chart). But for finding the right search results I don't think a table is the best options for the constraints I already mentioned!

[ruiwanguk]

@jadianes Thank you for the nice suggestions. Since it is relatively easy to experiment with the list layout , it offers the potential for much better usability, and we had experience with this on designing PRIDE Archive web, I would vote for trying out the list layout.

[jadianes]

@jgriss @ruiwanguk @noedelta I've just committed some changes to the search results page. Once @noedelta have them deployed, we can continue the discussion

[noedelta]

@ruiwanguk @jgriss @jadianes re-deployment done http://wwwdev.ebi.ac.uk/pride/cluster/#/list?q=&page=1&size=10

[ruiwanguk]

@noedelta @jadianes Thank you very much! I personally think the list layout is much better.

[jadianes]

@ruiwanguk @jgriss @noedelta I've committed new updates. A list of what has been incorporated so far:

• Peptide sequences should be in a different colour instead of white
• Change the colour that represents the results
• Change “Spectra” to “#Spectra"
• Merge "Prec. M/Z" and “Prec. Charge” into “ Precursor, and separate the values using “/"
• Precursor charge should be integer, this might need to be changed in the index as well
• Add “#Projects” to show the number of projects
• View as List is linking to the home page at the moment
• Breadcrumb needs to be changed to “PRIDE > PRIDE Cluster > Peptide Search"
• The three peptide sequence examples under the search box can not find any results
• It would be really nice to show number of species and number of modifications as well

I like @jgriss suggestion about incorporating precursor values into the title by the sequence. Should we try this and see how it looks like?

@noedelta @ruiwanguk right now the WS gives back zeros for most of the counts but numberOfSpectra.

[ruiwanguk]

@jadianes great stuff, thanks a lot.

I think it is worth a try on @jgriss 's suggestion.

About the zero values, we have found the problem, the values are available in the index, it is just the web service has been pick them up yet, so my bad.

[ruiwanguk]

ok, I am happy to announce that the numbers are back :1234:

[jgriss]

@jadianes @ruiwanguk @noedelta Thanks a lot for the work Jose! Personally, for me it's easier to quickly get the information I'm looking for with the new table view. Probably, it's just because I'm used to looking at tables. At the same time, the core information I'm looking for (cluster size, precursor m/z, charge state, modifications) are easier to compare that way since I just have to keep looking at the one column. In the other view I had to look downward and to the side to compare these numbers. Personally, I believe that most users will look for clusters of interest in the same way (largest clusters with a given precursor m/z / charge state and a defined set of modifications).

One additional comment:

• Add page movement controls (next, prev) below list as well.
• To me it would be nice to see the total number of pages. But this is a personal opinion.

[jadianes]

The latest committed code should include additional details on each result summary - once we redeploy we can check if the are OK.

[jgriss]

@jadianes @noedelta @ruiwanguk I am not sure whether the currently deployed version contains all of Jose's changes. So ignore if that has not been updated yet:

• [ ] At the moment the precursor m/z value and the charge state are missing in the search results.

[jadianes]

@noedelta @ruiwanguk @jgriss I'm going to add it back together with going back to the Frontier CSS only. Should I add the precursor/charge section again to the cluster summary page?

It would be great if we can deploy this today :)

[jadianes]

I've added a partial fix - couldn't make it work in the cluster summary page for some reason... My changes are committed

[noedelta]

Hi

I added a other small fix for having this working in the primary peptide (remove a pair of brackets the expression looks like work) however now it is broken in the unique peptides table :(

It is deploy,

Noe

[jadianes]

Fixed now.