Bob with Bioagents dialogue system

Bob with Bioagents is a machine partner you can chat with about molecular biology and solve problems together. The system can be launched with two different interfaces, CLiC and SBGNViz below.

To launch the dialogue system, please enter your name and email address, click on one of the launch buttons below, and wait (CLiC: 30 seconds, SBGNViz: 60 seconds) to get a dedicated instance of the dialogue system. Once started, the dialogue session will appear on the page and you can start talking with the machine agent. When you are done, click the End Session button above the dialogue interface.

The dialogue system was last updated on July 19 2020.

The CLiC interface focuses on text-based dialogue with an integrated tab for uploading and discussing lists of genes and experimental data.

The SBGNViz interface contains a large and interactive model canvas along with the text-based dialogue, allowing richer model building, analysis, and export.

Tutorial Videos

Explaining a perturbation-readout experiment In this video, we formulate a hypothesis for "How does SB525334 decrease the amount of SMURF2?" (a hypothetical experimental observation) by talking with the machine partner, which turns your hypothesis into a mathematical model of biochemical mechansms over the course of the dialogue, and is able to verify that the model reproduces the observation of interest. This video uses the SBGNViz interface. Link: Video

Uploading and discussing gene lists and experimental data to formulate hypotheses In this video, we upload a list of genes and experimental data and ask questions about groups of genes, and perform gene set enrichment analysis. This video uses the CLiC interface. Link: Video

Demo videos

Discussing COVID-19 gene expression data This video shows an example of uploading a gene expression data set, discussing pathway enrichment, relevant interactions, and possible drug candidates using the CLiC interface. Link: Video

Modeling and simulating a model of cancer drug resistance This video shows an example of modeling a hypothesized mechanisms by which cancer cells evade targeted therapies. Link: Video

Cogent: introducing the dialogue platform on which Bob is built This video introduces Cogent, a generic framework for developing dialogue systems for collaborative tasks, which is also the basis of this system. Link: Video

Talking with Bob

Assume you want to explain an experimental observation, or get some ideas for your next experiment. You can talk with the machine agent in some of the following ways:

  • Ask about drugs, transcription factors, miRNAs and their targets
  • Ask for mechanisms connecting molecular entities
  • Define an observation you would like to find an explanation for
  • Build up a model of a hypothesis by describing it in natural language
  • Ask about dynamical properties of the model being discussed

Both interfaces show example sentences you can say. While the system aims to support substantial linguistic variation, following the pattern of these sentences is in some cases more reliable.

The development of this system was funded under the DARPA Communicating with Computers program. The system was developed by a consortium of teams from Harvard Medical School, OHSU, Tufts, SIFT and IHMC.

Point of contact: Benjamin M. Gyori (benjamin_gyori [AT] hms.harvard.edu)

Privacy policy: The entered name and email address, as well as logs of the dialogues are stored. We use this information internally to improve the system and to produce usage statistics for reporting to the funding agency.