Sorting still not what a human expects: needs to sort by *inferred* number, not just by string.

[fragalysis-reporter]

• Version: 0.8.8
• Name: Frank von Delft
• Email: frank.vondelft@sgc.ox.ac.uk
• Description: See screenshot - x2562 should go above x10876 - that's what a human would expect.
  Should be quite simple to build a bit of smarts into the sort function?
• Reducers: https://github.com/m2ms/fragalysis-assets/blob/master/Reducers-1596040145158645329.json?raw=false

[boriskovar-m2ms]

While browsing live system I found out that this names varies in their format. Somewhere it's just X0107 with some suffix X0107:MAK-UNK-6435E6C2-8 somewhere it's ATAD2A-X1712_1, somewhere it's X0904_1 and X0904_2 where I assume that X0904_1 should go before X0904_2, somewhere it's NUDT7A_CRUDE-X1183_2 so there is a wide variety of patterns with variety of separators. Yes it can be done by searching for X and then trying to construct number and after _ (if exists) there will be secondary sorting criterion but it may has significant influence on sorting performance if many molecules are present on the selected site/sites. There is a risk that after implementing this it will become unusable. There are three options:

1. Always parse the number/s on the fly when sorting is invoked
2. When the data comes to front end I will parse the names and store the numbers which may slow down already slow loading process
3. Number/s will be extracted at the server side and sent alongside the other data which we are already receiving from the backend

[ak-m2ms]

Solution in branch #424

[boriskovar-m2ms]

The current behavior is that seemingly for random molecule the L and P is selected. See picture bellow:

The default and desired behavior for default selection when arriving on the page from landing page through selecting target is that first site is selected and L and P for the very first molecule are selected. Like on the picture bellow:

Bonus point: please sort the molecule sites by id ascending.

[boriskovar-m2ms]

The same bug is still present. To be more precise: The desired behavior is that L and P are selected for very first molecule that appears in the list in the UI.

[boriskovar-m2ms]

Please disregard the previous post. It's seems that mistake was made on my side and feature works as intended.

[boriskovar-m2ms]

Little bit of info about the sorting:
We expect the molecule name to be in the format of (aaaaaaa)Xdddd(dddd)(aaaaaaaaa) where 'a' is a character, X is delimiter or better to say it's prefix of the number part, 'd' is a digit, '' is a prefix of second sorting criterion and everything in between the parentheses is optional. 'Xdddd' is forming first sorting criterion and '(_dddd)' is second sorting criterion so X0001_01 goes before X0001_02.

[teriraj]

sorting works properly.

[phraenquex]

Sorting works properly, thanks!

There's a new bug, though: changing sort order (in the "Left filter") now takes so long that firefox tells me "a web page is slowing down your browser" - literally minutes. I'll have to raise a new ticket.

[phraenquex]

After reviewing with Rachael: we'll raise a new ticket about the sort/filter modal, and won't move this to production until that one is fixed too.