rnacentral/rnacentral-sequence-search

RNAcentral sequence search cloud infrastructure

rnacentral-sequence-search's Issues

Show how long the search took

If you don't measure it, you can't improve it!

Show contact us link when jobs fail

Would be better to give users the option to contact us when all job chunks fail. Maybe a contact us link would be enough.

Accept fasta file as an input to job submission

• Accept .fasta file on input (do most of the validation on the backend)
• Accept DNA sequences
• Don't accept random characters (probably, validation on frontend and backend sides)

Show the exact match using md5 lookup + highlight it when the results are ready

The JS code can be ported from the current website.

Facets ordering like in the main website

Heartbeat, failure recovery

Jenkins one-click deployment

• add 'quick' parameter
• create an NFS server installation instead of volumes

Make job submission browsable in api (probably, through creating a ModelViewSet)

Integrate frontend components with test.rnacentral.org

Top priority!

Montior the state of consumers from producer

If a consumer is stuck in 'busy' state and no job_chunk is associated with that consumer (recently?) - restart that consumer from producer with ansible.

Accept DNA

Should accept a DNA input as well as RNA.

Handling partial_success

• redirect to results even if they are incomplete
• display warning that results are incomplete/timeout

Search results take a very long time and then facets don't load

We may need to limit the maximum number of hits we display using facets.

http://193.62.55.80:8002/result/9afd1c9f-4a93-4fcd-bc61-41b2d4c6eb4a

Remove unused columns from the database

"Service discovery": consumer should save its coordinates to the database upon startup

nhmmer crash for one of the chunks

Permanently stop that chunk, continue processing the test case, but leave it with incomplete status.

How to test: use a wrong command option.

Limit the list of facets to display with a fixed length

Add consumer ports to the database

Probably move the default consumer port to 8001.

Make sure that request, cancelled due to closed browser tab, is handled gracefully

Display a warning about malfunctioning text search on frontend

job_ids should be hashes instead of integer autoincrements

rna_type facet click causes a text search crash

Try increasing the number of elements in a facet to ~100 and make them scrollable

Debug the entries filtering in facets

When we click on a facet, it filters the facets themselves, but not the search results.

• email Nicola
• add a limit on total number of hits (e.g. 10k)
• paginate sequence search results (I'm currently getting all of them)
• don't post to ebi_search_proxy every single time, if the search was already done?
• if text search is not available - add a flag to the output json and make a graceful degradation

Cloud deployment testing

Sorting order by e-value doesn't work

Infinite scroll doesn't work in Safari

Add counts to the facets as in the main RNAcentral website

Fix weird database names in frontend interface

Hide names of chunks being searched

I fear that users will start wondering about what is in whitespace.fasta and so forth. Hide it and use something simple.

Input validation

It accepts an input of just >, which is invalid.

Something's wrong with nhmmer_parse - gap_count and gaps are always 0

Different results than in current search

When I search for:

GGCAGUUAUUAGGUUCAGUCCUGGAUCAAUCCUCGCUGAUGUAAAUGUAACAUUUACAACUGUACCCACUAAUGAGGACAUAAUCAGCACUAUCACCACUGCAAACCUUACUUUUCCUGUCACCGGAGCAGAGGGUAAGUUAAACGAGCAUCUUCAACAAUGCAGUAUCAUUUUAAAAAAUUUUGGAUGCUUUUGGUAUUUCUCCUGUGUUAUUAGACUAUACAUUAGCUGUUACAUGUGA

I get results in current search but none in the test site. Is this because of the subset searched?

• Should update the database files with the later version from:

ftp://ftp.ebi.ac.uk/pub/databases/RNAcentral/current_release/sequences/.internal/sequence-search

Fix restart of consumer and producer processes.

Make database directory location customizable

Ensure database consistency with transaction isolation

Often we do 2 or more SQL queries in the same view. If second query fails, ideally we shall rollback the first query to keep the DB in a consistent state. Should we introduce transaction isolation?

Set timeouts on nhmmer jobs

If one of the jobs times out:

• kill that chunk
• set status of the whole job to incomplete
• display a warning in the UI

How to test:

• manually set a wrong starting time in the database

Improve formatting of alignment statistics table

• https://github.com/s-yadav/react-number-format
• https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/NumberFormat
• https://www.carlrippon.com/formatting-dates-and-numbers-in-react/

Facade for searching specific databases

Producer web interface and REST api by default should search against all-except-rrna*.fasta and whitelist-rrna-*.fasta.

Consumers could still work the same way they do.

Cross-browser testing

When zero results give option to search all data

As I understand it right now we search a subset. If we do a search that gets 0 results, then we would probably want to show a link to redo search with on all sequences.

Make consumer and producer same python package so that we don't duplicate the code

Swagger api documentations, correct examples

Fix proxy, make sure that facets work

Infinite scroll of results

https://github.com/CassetteRocks/react-infinite-scroller#readme
https://alligator.io/react/react-infinite-scroll/

Make DevServer and HtmlWebpackPlugin work together and serve statics from RAM

Currently webpack dev server is serving the bundle from the filesystem, not from RAM, which is making development process insane.

Show what was searched in results page

Generally I forget what I was doing.

Make all the status constants Enums

Otherwise a typo error or re-design might break the project.

Make Swagger api a built-in frame within React frontend

Generate ansible hosts with Terraform module