Biomart Vs Intermine Vs Atlas

Question

[Andrea_Bio]

Hello

I was hoping people with experience of using biomart/intermine/atlas could tell me some of the advantages and disadvantages of the different tools.

I am looking to find a tool that will provide a ready-made interface for browsing the underlying data source and also has tools for inserting/updating data.

I only have experience of biomart and I know there is no api for data inserts/updates. Intermine has java and perl apis but i don't know if there is anything in these for inserts/updates.

What are the major differences between biomart and intermine? When would you use one over the other? Biomart seems the most ubiquitous.

I have come across ATLAS in my searches but it doesn't seem to be in widespread use. Is this correct?

ANother alternative would be to use middleware tools on top of a CHADO database as discussed here. Or to use Djnago with chado

Thanks in advance for your input

Question edit:

I am already biomart rc5 as a virtual mart on top of the a normalised data source and performance is ok. The virtual mart automatically detects inserts/updates. As the answers given suggest, biomart is for read-only data. I was hoping to use a biomart front-end (because people are familiar with it) for a changing dataset. One option is to use a normal database with ORM tools for modifying the data with a virtual mart on top. ANy comments on this approach welcome.

[Chris Evelo]

Which Atlas do you mean? There seem to be severall genomic tools called Atlas. Including Genome Atlas -: http://www.cbs.dtu.dk/services/GenomeAtlas/ and ArrayExpress Atlas (which is really for expression data)

[Chris Evelo]

I deleted my answer since that was really more about the content of the major implementations of BioMart and Intermine and thus lead to a discussion that did not really answer your question.

Answer 1

[Michael Schubert]

The goal of both BioMart and Intermine is to increase query performance on read-only datasources. This means that you usually:

• have a normalized database (SQL-compatible with BioMart, XML schemata with Intermine) with all your data already in place
• use that to build query-optimized de-normalized tables plus the web-frontend

The denormalized database is ill-suited for the insertion of new data. What you do in that case is update the data in your normalized DB and then build BioMart/Intermine anew. What you could do is automize this process using scripts.

For updating data in your original database you might want to look into phpMyAdmin or similar database frontends.

Considering ORM-only tools like Hibernate or Chado, the problem is that you still have to write the web interface yourself. GMOD on the other hand might offer some advanced solutions as well (but I don't know enough about it to tell you what they are).

[Joachim]

The performance is the same as when you query your DB manually with SQL. There is no performance penalty for not materialising a mart, but only a performance improvement if you choose to materialise it.

Inserting or updating data in a materialised mart might / might not be easy. It really depends what you try to do and that is why there is no general implementation for carrying out inserts/updates on a materialised mart yet.

[Alastair Kerr]

I've just started looked a 0.8rc5 of biomart. http://www.biomart.org/rc5_documentation.pdf It looks like it is now possible to attach any relational database and combine that with known marts into a non-materialised 'virtual mart'. No idea what the performance is like but I should know in a few weeks.

[Andrea_Bio]

HI Joachim - i second that comment. My virtual mart is fine for speed. Even a mart sitting on top of the ensembl server is fast.

[Radek]

InterMine does provide a web interface and an API in languages like Perl, Java, Python, Ruby, JavaScript or Node.js. As for GMOD: InterMine FlyMine, RGD RatMine, SGD YeastMine, modENCODE modMine and others use it, be it fully or through embedding.