Humboldt-Universität zu Berlin - Mathematisch-Naturwissenschaftliche Fakultät - Wissensmanagement in der Bioinformatik

Frequently asked questions


Installation and system requirements

Which Java version does Ali Baba require?
Please use Java 1.5 or higher to ensure compatibility.
Why does Ali Baba not start on my system?
Please ask your system administrator to install Java Web Start (javaws) on your machine. For the first installation, administrator privileges may be necessary.
Do I always have to use a web browser? Is there a possibility to link Ali Baba directly from my desktop?
When launching Ali Baba Java Web Start asks if you like to create a shortcut on your desktop. Next time you can start Ali Baba without a web browser by simply clicking the shortcut item.

Large queries, large graphs, speed

My system slows down if graphs get too complex.
Try setting [Filter preferences] from the [Preferences] menu before or after submitting your query. Also, high quality rendering and animated graph layout take a lot of computation time. Reducing render quality or switching off layout animation in menu [Preferences] will help.
Ali Baba has a path length filter that reduces a graph to the selected node and its immediate neighbors.
My graphs are just too large and complex. There are edges between almost every pair of objects.
Try using the "Confidence" slider on the bottom. Ali Baba assigns a confidence to every relation between objects. The higher the score, the less likely is the relation to be a 'random' co-occurrence.
Most often, the radial layout provides less overloaded screens. Toggle it on using [View|Layout|Radial] from the menu.
In addition, Ali Baba has a path length filter that reduces a graph to the selected node and its immediate neighbors.
I want to query for many citations (500+). The progress bar goes up until 95% and then stops.
For large queries (500 and more abstracts), you might want to try versions of Ali Baba that use a maximum of 1.0 GB or even 2.6 GB of memory instead of the lower default setting. Click on the size to start the application.
Why is the search limited to a maximum number of 1000 PubMed abstracts?
Restricting the maximum number of PubMed abstracts helps to save server-side processing and network transmission time of result sets. This limit might increase in future versions.

How can I find particular objects?

How to find a particular entity among the many in my result set?
Use the input field on the bottom to search for nodes by an entered keyword. Nodes with labels matching the keyword become highlighted. By default, nodes get highlighted as you type; to toggle this feature off disable [Preferences|Incremental search in graph].

Missing nodes and edges

Why does my result set contain only few entities and relations?
Check your filter settings in the [Preferences] menu. AliBaba remembers your last choices that might hide entities and relations below a confidence threshold.
I know that a particular object I am studying is associated with another one. This information is not shown in the graph, despite both objects are present.
Ali Baba does not use any annotations from external sources. The data is collected on-the-fly from PubMed abstracts only. If a relation is not discussed in any of the abstracts, it cannot be extracted. Try altering your query, or construct your graph from multiple queries by switching the 'append' modus on.
I miss some relations (edges) between objects that should be connected.
Ali Baba does not visualize relations between every possible pair of objects types. It currently searches for protein-protein interactions, (sub)cellular locations of proteins, proteins discussed with species or diseases, mentions of drugs together with diseases, and relations between diseases and tissues. However, there might be some more in the near future. If the type of relation you need is included in this list, Ali Baba might not find a relation for two reasons. One, the objects might not be discussed directly together in the PubMed abstracts selected by your query. Then, the relation might be recognized as too vague according to the textual evidence. Try decreasing the confidence score in this case.
Some words are marked with the wrong entity type.
This may have two reasons: i) When we found more than one possible entity type for a word ("Nf2" could be either a gene/protein or a disease), separate nodes for all these different entities will appear in the graph; however, in the text, we use only one color to highlight the name - this might not be the one you expected. ii) Ali Baba misinterpreted the word for a wrong entity type; if you see such an error, please help us using the feedback modus! Simply highlight the wrong word, select the correct type from the appearing dialog, and submit your suggestion.
I need types of relations that are currently not visualized by Ali Baba.
Are there going to be more types of relations?
We plan to include data on enzyme kinetics (enzymes, compounds, reactions, kinetic constants) and associations of genes with diseases/disorders in the near future. In addition, there will be annotations of proteins/genes with Gene Ontology categories.
I need other types of objects for my queries. Are there going to be some more entity classes?
We plan to include recognition of enyzmes and compounds in the near future, as well as Gene Ontology categories (biological process, molecular function, and cellular component (cells already get extracted)).

Error messages

I get an error message: No Results. Possible Reason: Read timeout.
Try reducing the maximum number of hits for your query. If that does not help wait a little and try again. Current network traffic might be too high.
I get an error message: Could not read the result set!
This is propably caused by a server-side processing failure, e.g. an annotation service is currently not available. Please try again later.

Submitting comments and suggestions; reporting bugs

I would like to report a bug, or have a suggestion or comment
We appreciate any comments, suggestions, or bug reports on Ali Baba. Please send your comments to AR (rheinlae youknowwhat informatik dot hu minus berlin dot de) and UL (leser youknowwhat informatik dot hu minus berlin dot de). Or, in case you spot entity names marked with the wrong type (for instance a word that is marked as a protein, but really is a cell type), please use our feedback modus.

Please also see the section on known bugs.