InterviewNumber2

Notes from Interview with researcher at Health Science Center

15th February, 2006

Interviewers: Chris Workman, Allan Kuchinsky

Usage of Cytoscape

She is working on large scale data on viability using RNAi knockdowns in tissue cultures, looking for viability after treatment. She uses an expression matrix of genes to look for things that respond to treatment in a similar way and if they are connected in the network. She also looks for other proteins that interact with the hits. She uses the network downloaded from BIND (BINDhuman) and would like to download from DIP. She would like to combine information from different database sources into larger network. At the retreat she suggested a number of desired features for the future, got a very good response. She said that if the developers act upon what was discussed at the retreat, she would be very satisfied.

Undo

Undo in 2.2 sort of gives a false advertisement, all it does is undo node shape. She would like more comprehensive functionality, such as the ability to undo the seeding of node and edge with random colors, i.e. the ability to revert back to the previous settings.

Visual Styles

She spends most of her time in Cytoscape setting visual attributes. She spends a lot of time struggling with the visual mapper to map attributes related to RNAi knockdowns. Most frustrating is the behavior of the system when switching between visual styles. If she creates a new map style or wants to go back and change one later, then instead of pulling up the default calculator, Cytoscape pulls up the most recently edited mappings. For example, with edge line type, if she wants to change all lines to 6 thickness in several different visual styles, then switches styles, the vizmapper does not automatically update when the styles are switched. The edges in the network view show the edges with the appropriate thickness for that visual style, but the vizmapper configuration window is not automatically updated. Instantaneous update on edits to mappings and styles, e.g. instantaneously recoloring nodes in the displayed network when node color mapping is changed, would be nice, but not essential. The ‘Apply to Network’ functionality is just fine. She found the feature of seeding with random colors to be potentially useful, but very frequently the choice of color sets was not good. We might want to consider keeping a set of ‘color schemes’ and have a ‘seed with random color scheme’ functionality. Some aspects of the vizmapper are very clunky, for example going down through levels of definition. Also, sometime the vizmapper window doesn’t resize correctly. The GIFs provided with the linestyles are useful. She doesn’t currently use edge arrow types.

Node IDs

The mapping of synonyms is a major problem, e.g. mapping from DIP names to names that Cytoscape can recognize. Also, she wonders if Cytoscape could automate some of the process of getting information from databases on the Internet. That would be very helpful in making sure that they have up-to-date versions of the database. People in her group have asked: “Why can’t you just connect to the Internet and have it pull down the latest networks for you?” Another very useful functionality would be using Cytoscape to merge different public networks instead of doing this before loading into Cytoscape.

Cytoscape Editor

She likes the ability to modify the network some. She uses the Cytoscape Editor to add interactions to existing networks. She said the editor was fine, works easily.

Installation

She is on a Mac, installation was just fine. She noticed a performance improvement in 2.2. Largest network she has is just under 9000 nodes She can usually do analysis on the network without having to generate a view. She is definitely interested in what is happening with the new renderer and when that will be available.

Plugins

It was easy to download plugins, drop them into a folder, then run them. She uses MCode, JActiveModules, graph merge, messed around with cPath. She will start using Bingo, but the set of GO annotations for drosophila is poor now. Many plugins are not intuitive and she only understands the functionality of the plugin after she reads the paper. User guides would be helpful.

Filters

She uses filters a lot to generate subnets: she uses numeric and topology filters. For example, she has a filter that finds nodes with numbers of immediate neighbors over a threshold value. She likes the modifications to topology filters since version 2.2. She uses String filters to find specific nodes, but thinks that it is too much overhead. She would like ‘Find Node’ and ‘Find Edge’ commands (with control-F accelerator).

Cytopanels

She uses CytoPanel 2 a lot, and likes it a lot. She doesn’t ever use cytopanel 3. She finds that CytoPanels cramp the desktop a bit, so she undocks CytoPanels and brings them up when desired.

General

More frequent releases would be better, perhaps a patch version every one-three months.

Meta-comment on providing feedback

Cytoscape-announce is pretty good about keeping her up to date Should she report usability issues in mantis or is there a better venue for this? Where should she report things? Interestingly, both she and Nazilla Miller spoke about being hesitant to bug the developers too much. We reassured both of them that the developers were quite interested getting their feedback and learning about what problems they may be having.

Both of us were struck by the fact both of the people we interviewed had actually read the help documents and user manuals and said that they needed more. The concern is that many users may not have as much time (or patience) to read through these help materials. Nevertheless, it seems that having thorough documentation is essential, in particular documentation that is framed in terms of the problems they are trying to solve using Cytoscape.

It also seems that the notion of having Cytoscape ‘modes’ that are tailored to specific use cases, e.g. gene expression analysis, along with documentation tailored to the use case, might go a long way towards helping the non-programming biologist who may not have the time or inclination to download plugins.