Frequently Asked Questions
Below are some of the frequently asked questions we get from our clients. If you have more questions, or would like clarifications, please feel free to contact us.
I work in R and Python. Can I use these with your system?
Yes, we have SDKs that allow you to plug in your scripts and models.
I like my processes. Why should I switch to yours?
Our platform is based on independent nodes that gives you the flexibility to build whatever you want. Our aim is to amplify your current processes, not replace them.
How Tag.bio Compares
How are you different from Tableau and Qlikview?
Tableau and Qlikview interact with users through dashboards. These dashboards show individual answers to pre-selected questions. They provide snapshots, but they do not enable investigations.
Tag.bio allows data scientists to build analysis apps that guide users through their investigations by asking their own questions. Their answers are saved into a history that is reproducible, replayable, and shareable.
Our advantage over those platforms is that our user experience is driven from within each data node, and therefore each data node can provide its own native, domain-specific suite of functionality.
- For a life sciences example: The Pan Cancer data node has an analysis app on mutation among other apps related to the node.
- For a healthcare example: The hospital data node has an analysis app on 30-day-readmission among other apps related to the node.
How are you different from R and Python?
R and Python are coding languages, but not a platform.
Tag.bio enables data scientists to integrate R and Python into any data node, turning their scripts into analysis apps. Publishing an analysis app allows the data scientist to present the domain experts with a parameterizable data product. This reduces the number of requests that the data scientists get for parameter changes. The data scientists can review how the apps are experienced and iterate to come up with the best solution.
For the domain experts, the apps are available on the Tag.bio portal where they can run the scripts as they would run any other app. They can build a cohort, send that cohort to the script which runs automatically, then display the result into an appropriate visualization. This allows them to review and extract useful data artifacts, all of which are saved in their history, allowing them to replay and share the results.
How are you different from Jupyter Notebook?
Jupyter Notebook is an interface for data scientists, but not for domain experts.
Tag.bio provides a way for the data scientists to make their scripts and models usable by the domain experts. The domain experts also have a user interface where they can iterate on the parameters by themselves until they find some useful data artifacts.
What data types can you take in?
We take in analysis ready data, such as metadata that has been derived from a process that transformed data.
For a life sciences example, there’s DNA sequencing: We would use the adapted variant call files (VCF) to analyze descriptive information, but we don’t store the images.
For healthcare, there’s a radiology images example: We analyze descriptive data of the images, but we don’t store the images themselves.
Do you clean data?
We can perform standard data management such as uniform datetime stamps, numeric normalizations, mapping to common concepts. If you require extensive data cleaning, we can accommodate that in collaboration with trusted service partners.
How does my data get into a data node?
Our platform has a data mapping/configuration layer which loads data from tabular data sources (e.g. CSV, SQL), and can either instantly load data as-is, or perform sophisticated joins and transforms to produce greater value.
How long does it take you to create a node?
Typically, the process takes hours to a couple of days depending on data volume, data types, and the readiness of data.
How long does it take to make a new analysis app?
Typically, the process takes thirty minutes to a couple of hours depending on the complexity of the questions. For example, apps that provide a summary of data take thirty minutes. Apps that use novel algorithms and visualizations can take hours to days.