wormbase / wormicloud Goto Github PK

Now the default is set to case sensitive, but if somebody tries to do a search on tpc without setting the 'case sensitive' option, they find many more results than the references in the cloud, and this may be confusing.

Pubmed searches are on abstracts only, but they can retrieve papers not yet in textpresso.

Add link from gene name cloud

the current word tokenizer treats dashes as word delimiters and gene names are being split (e.g., daf-2)

Right now if the user enters more than one word in a single query box (e.g., 'positive' and 'result') the search is performed with the default lucene engine, which searches all documents containing 'positive' OR 'result'. By adding double quotes we want to force lucene to search for the sentence 'positive result' (since the two words are entered in the same box and not in separate ones).

As suggested by @chris-grove, we should find a better color palette to avoid using similar colors for different words or clusters. As I was mentioning at today's meeting, we are already using a package specifically designed for generating distinct colors (https://www.npmjs.com/package/distinct-colors) for word trends, but it doesn't really work as expected. One possible solution would be to handpick a list of orthogonal colors and always use this list without shuffling the colors every time. This would also provide stable results in terms of color scheme and could help users identify the results quickly.

These are not in any categories in Textpresso, but some genes don't have a public name yet and sequences would improve the cloud, especially for protein-protein interaction analysis. We need to ask Michael if they can be easily added to textpresso.

Suggested by @chris-grove. This can't be done in the React word cloud component that we are currently using (https://www.npmjs.com/package/react-wordcloud), but it could be part of an alternative graph/bubble display. Here are some ideas:

https://observablehq.com/@d3/clustered-bubbles
https://github.com/crubier/react-graph-vis

This could end up in a tab-based view - view word cloud, view word trends

The idea is to enter a person name and get a wordcloud of his/her abstracts. Can also be extended to more than one person.

Provide the following options from the UI:

• papers mentioning both genes: use TextpressoAPI to get fulltext of all papers where both genes are mentioned (document search)

Use abstract of matching papers (up to 200-600) to generate the word cloud

• show the counter in a new column
• sort by number of curated entities
• get the data from the WB API - tree display in WB website

Add table with paper references at the bottom of the page.

Show title and PMID and year, sorted by year

@WBjae, can you add a list of words to be removed from the cloud?

To enable linkouts from WormBase interaction pages

Consider both directions when pulling the data from the DB.

Right now the minus sign always removes the very last box. We should place a minus sign after each box to let the user remove specific boxes instead of the last one.

• Put cluter id -> all keywords in cluster
• Check numbering of clusters

Textpresso categories queries are too slow, and pap_gene already contains a lot of gene names. Future automated extraction pipelines should make pap_gene contain gene names matching those extracted by tpc

We could add a tooltip to the words in the cloud with a "show me references containing this word" link

functions to remove a single word from the cloud. Having a popup window with a 'remove' button from the cloud would be ideal