13. Analytics and Charts

Mosaic allows users to generate charts for attributes at both the project and collection levels. This is particularly useful for data quality control across all projects, and for understanding the composition of a cohort at a glance.

13.1. Charts at the Project and Collection Level

Charts can be created from the Analytics page (reached from the left menu) in both projects and collections. Because a chart summarizes the values of an attribute, the level at which you create it determines what it shows:

• In a project, a chart summarizes the values recorded for that single case — useful, for example, for reviewing sample-level quality metrics.
• In a collection, a chart summarizes the attribute across every constituent project, turning the cohort's data into a single picture. This is where charting is most powerful: one chart can show, for example, the distribution of Diagnostic Status across all cases, the spread of sample sex, or the range of a quality-control metric for the whole cohort.

Certain charts are displayed by default, however, default charts can be changed using templates.

13.2. Creating a Chart

On the Analytics page, use the Actions button to create a new chart. Charting works from the attributes recorded in the project or collection, so you choose the attribute to chart and the chart type appropriate to its data — for example, a distribution of a categorical attribute such as diagnostic status, or a plot of a numeric attribute such as read coverage. Charts therefore depend on attributes being populated; see Project Attributes and Sample Attributes.

13.3. Pinning Charts to the Dashboard

Charts can be pinned to the Project Home or Collection Home, where they appear as cards. Pinning the most informative charts — for example, the cohort's diagnostic-status breakdown. Pinning follows the same pattern as pinning attributes and conversations, described in Project Attributes.

13.4. Charts for Quality Control

A common use of analytics is monitoring data quality across a cohort. By charting sample-level quality attributes at the collection level, you can quickly spot outliers — for example, a sample with unusually low coverage, or a sex-check result that does not match the recorded sex — and follow up on the specific case.