• Give user hint when they specify their input prefix

  1. syntax
  2. example

• When user submit their query, check whether their input value actually match the

currently, links in the "Action Log" box load in the current window. Because biothings explorer does not save state, the current session is lost when you return. Better to load the link in a new tab.

MWE: http://biothings.io/explorer/api/v2/directinput2output?input_prefix=go&output_prefix=goDescription&input_value=0042535

Got json['data'][0]['output']['object']['id']:

"monocytes during acute inflammation and which is responsible for a diverse range of signaling events within cells, leading to necrosis or apoptosis."

Expecting:

"goDescription: Any process that activates or increases the frequency, rate or extent of the chemical reactions and pathways resulting in the formation of tumor necrosis factor, an inflammatory cytokine produced by macrophages/monocytes during acute inflammation and which is responsible for a diverse range of signaling events within cells, leading to necrosis or apoptosis. "

The part preceding the forward slash in front of "monocytes" is truncated.

Is it because of JSON-LD? Thanks!

e.g. {'input': ncbigene, 'output': uniprot, max_no_api: 4}.
This path couldn't be displayed on the sankey plot.
But if we change the max_no_api to 3, it can be displayed.

Looks like an issue with the plotly side. But needs to be further confirmed.

e.g. humanbase API requires two input parameters -- tissue and gene.
In this case, if only one parameter is specified, the interface should pop out a window reminding the user to provide another parameter.

For example, clicking the "1017" node multiple times results in this star pattern. Instead, should add a popup that just says "all endpoints are already shown".

Set up crawler to fetch the databases and version (if they provide) used in those APIs, e.g. BioThings APIs and BioLink APIs, Pharos, DGIdb, etc.

BioLink: https://monarchinitiative.org/about/sources
MyGene: http://mygene.info/v3/metadata
MyChem: http://mychem.info/v1/metadata/
DGIdb: http://dgidb.org/api/v2/interaction_sources.json
Pharos: https://pharos.nih.gov/idg/help
EBI OLS: https://www.ebi.ac.uk/ols/api/ontologies?page=0&size=210
MyDisease: manually curate one
ChEMBL: https://www.ebi.ac.uk/chembl/faq#faq33
PubChem: https://pubchem.ncbi.nlm.nih.gov/sources

http://www.obofoundry.org/ontology/mondo.html
MONDO has a JSON dump which is easy to parse and put into an API.

1. Reactome
2. Pharos
3. Chembl (more endpoints)
4. GeneProf
5. NCBI eUtils (more endpoints)

Given the state shown above, I would expect that unchecking the box for WP179 would remove the edge from my network. It does not however (and even adds another edge between the two nodes). Either the behavior should be changed so that unchecking the box removes it from the network, or the checkbox should be replaced by a simple "add to network" link/icon.

1. Add buttons on the cytoscape/plotly graph to enable users download those graph in PNG or jpg format
2. Add buttons on the graph to enable users directly download the data, e.g. in triples.

https://api.varsome.com/lookup/10190091015942290001

https://github.com/common-workflow-language/common-workflow-language

1. There should be one unique color representing API endpoints
2. Different colors should be used for bio-entities for different semantic groups.

Hi Kevin,

MWE: http://biothings.io/explorer/api/v2/directinput2output?input_prefix=do&output_prefix=hp&input_value=1498

The json["data"]["target"] field is like "hp:HP:0002090", which used to be "HP:0002090".

Similarly I am getting IDs like "HGNC:HGNC:11138" and "UBERON:UBERON:0002113". Could you please check it out?

Thanks!
Yao

e.g. input->ncbigene:1017; output -> drugname: riluzole
return all intermediate nodes which could potentially connect these two biological concepts.
e.g.
mitomycin c  -compound
fanc* (genes)

• web version

1. all edges should be shown
2. when clicking on one of the nodes or edges, the additional info should be displayed.

• API

1. Each path should be a list
2. Each element in the list should represent an edge
3. Need to find a way to populate this kind of structure

e.g. drug - uniprot ID contains multiple edge options, e.g. transport, enzyme, etc.
We should not only include the multiple edge info, but also increase the weight of the edge.
E.g. if there are 4 edge options, the weight of the sankey plot for that edge should be 4.

Add function to export Python code,
//COMMENT SHOWING HOW THE PATH IS GENERATED

1. PATH = [....]
2. import ...
3. get_output(PATH, _INPUT, ...)
   //COMMENT SHOWING HOW TO DO ANALYSIS WITH THE OUTPUT

https://europepmc.org/RestfulWebService#!/Europe32PMC32Articles32RESTful32API/references

Load CDK2 (entrez gene 1017), then click myvariant.info. biothings explorer reports no variants found. That's because it is querying this URL

http://myvariant.info/v1/query?q=clinvar.gene.id:1017%20OR%20emv.gene.id:1017%20OR%20dbsnp.gene.geneid

when it should be querying this URL

http://myvariant.info/v1/query?q=clinvar.gene.id:1017%20OR%20emv.gene.id:1017%20OR%20dbsnp.gene.geneid:1017

TODO

• Should put a table underneath the Sankey plot listing the endpoints, together with the location of API landing page, Smart-API specifications

• Migrate all Smart-API specifications to Smart-API repo

• Read the info to populate the table from ‘api_list.yml’ file

Endpoint has been set up to monitor the status of APIs: http://biothings.io/explorer/api/v2/apistatus
The status data is refreshed every hour.

Need to include this info in the front end.

BioThings Explorer has the capability to display input/output information for each API endpoint.
We should categorize APIs into two separate categories:

1. APIs which has metadata ready and can be called
2. APIs which has metadata ready but can not be called, due to
   a) They doesn't return JSON format, e.g. SQL, SPARQL query endpoints
   b) They accepts multiple parameters (currently we handle one parameter only)
   c) The return format cannot be JSON-LD labeled, e.g. pharos

Those API still should be discoverable using BioThings Explorer, but when making actual API call chains, they should be excluded.

TODO:
Each API endpoints in the BioThings Explorer registry should have a field called tag, which categorize their functionality.

Check why Biolink API endpoints couldn't be displayed on the options in 'explore endpoints' section.

just a step-by-step guide that shows one reasonable use case...

When the input is very long json file, the jsonld.biothings.io couldn't handle it.

User could specify to use "fast" mode, so only the results from API which responds fastest will return.

A new user who has figured out how to add a new node and then clicks that node will see this:

This layout is rather confusing. Apply some layout function to separate the nodes.

Current implementation of select doesn't contain any search function. This is not very user-friendly, when the input is large.
Try using select2 instead:
https://select2.org/getting-started/basic-usage

1. Highlights
2. Introduction

Set up crawler to fetch the databases and version (if they provide) used in those APIs, e.g. BioThings APIs and BioLink APIs, Pharos, DGIdb, etc.

BioLink: https://monarchinitiative.org/about/sources
MyGene: http://mygene.info/v3/metadata
MyChem: http://mychem.info/v1/metadata/
DGIdb: http://dgidb.org/api/v2/interaction_sources.json
Pharos: https://pharos.nih.gov/idg/help
EBI OLS: https://www.ebi.ac.uk/ols/api/ontologies?page=0&size=210
MyDisease: manually curate one
ChEMBL: https://www.ebi.ac.uk/chembl/faq#faq33
PubChem: https://pubchem.ncbi.nlm.nih.gov/sources

Hi Kevin,

MWE: http://biothings.io/explorer/api/v2/directinput2output?input_prefix=chembl.compound&output_prefix=pubchem.compound&input_value=CHEMBL503

There are json["data"]["output"]["object"]["id"] fields containing spaces, like "PUBCHEM.COMPOUND:SID: 26697338".

Is it on purpose or an error? Thanks!

Best,
Yao

https://www.ncbi.nlm.nih.gov/research/bionlp/litvar/api/v1/entity/litvar/rs121913527%23%23

Provide functions for users to filter the path based on relation(predicate) between input and output.
e.g. Drug, Protein may have multiple different relations, e.g. transport, enzyme, target, etc.
Users can select the path which specified drug -> targets -> Protein.

URIs currently not supported by http://identifiers.org ==> switch to biothings.io/explorer/vocab
URIs using biothings.io/explorer/vocab ==> add API endpoint accessing these
Just include some basic info in the response: e.g. name, preferred name, syntax, etc.

https://github.com/iVis-at-Bilkent/cytoscape.js-expand-collapse
When there are too many nodes on the cytoscape graph, we should explore how to expand or collapse these nodes.

Input data could have duplicates with the output data.
However, there shouldn't exists any duplicate items with in the input data or output data alone.

https://www.ncbi.nlm.nih.gov/CBBresearch/Lu/Demo/tmTools/RESTfulAPIs.html

somewhere near the search button, provide a selection of sample queries.

Internal function to convert JSON array to JSON object
By default, wrap the JSON array under 'data' field to form a JSON object.
Then create the JSON-LD context file based on the new JSON object.

• Merge based on the primary ID

• If one ID includes multiple entries from different APIs

1. Group the results by API

2. Display as a list

Add a function where user can define their own path.
This is to handle cases where there are duplicates in the paths,
e.g. gene -> pathway -> gene -> drug
Start with input,
-> Second option set include all APIs linked from that input
-> Third option set includes all outputs of the previously selected API
-> Repeat...

This starts getting in to heavier UI development, so we should discuss the game plan before committing to implementing this ticket. But, imagine the scenario below:

currently I can add all 14 of those WP pathways to the network, but they show up as separate nodes. Suppose I want to get all genes associated with any of those pathways. I would want to aggregate them into a single node. If I click that node, it should show me services that take that node type. If I click a service, then the explorer should execute 14 calls for each of those pathways (or even better, a single query that combines those 14 pathways).

it's confusing because it overwrites your network, but it's fundamentally a different thing than the network explorer. I think this should be on a completely separate page/tab.