Results Plot Revamped
STARInsight depicts the results of K-Means and Probabilistic Principal Component Analysis jobs on a scatter plot. From this plot, you can select outliers and save them as a new analysis set for further examination, or relabel data points based on project metadata. In this release of STARInsight, we've refactored the results plot for faster performance when zooming, scrolling around, or selecting samples on a plot that contains thousands of data points.
The same set of functionality as the prior version of the plot is available with this new results grid, but instead of having zoom buttons to the right of the plot, you can use standard mouse controls to closely examine clusters of tightly grouped data points. Hold the SHIFT key to select samples and save them as a new analysis set.
New Variant Exporter
STARInsight's variant repository tracks several pieces of genomic information including genotype, location (chromosome and position), read depth, and more. You can use the Variant Exporter app to output an Excel file with results data split across five tabs.
- Variants: a list of all the variants included in the exporter results. This list will be governed by the pre-processing filter set you selected when running the analysis.
- Genotype: a table where each column is a variant (Chromosome:Position), each row is an individual sample included in the export, and individual cells are the genotype values (0|0, 1|0, etc.) for each variant / sample combination.
- Quality Score: a table similar to the genotype table, but with each cell depicting the quality score for each variant / sample combo.
- Read Depth: a table like the Quality Score and Genotype tables, but which provides the read depth.
- Imputed: this table indicates whether a given variant / sample combination has been imputed in the filter set you have specified. Read here for more information on imputation.
If you prefer a simplified report which only includes genotype values, check the "genotype only" box when running the report.
Comments from our user base led to these little improvements...
- An unsaved changes warning appears on the Search page if you select a project metadata facet like "population name" and navigate away without saving a new analysis set.
- We swapped out a couple icons (Apps, Notebook) in the header for some brand new ones.
- You probably noticed that the UI sports a new color palette. We think this makes it easier to spot certain controls and distinguish other details, but tell us what you think.
- The new results plot includes grid lines to aid in comparison with axis values.
Things We Fixed
- The names of BED files no longer disappear from BED file widgets on the pre-processing page.
- We removed some misleading row and column counts from the variant exporter results page summary section.
- It is no longer necessary to save an unwanted filter set before deleting it.
- STARInsight now treats "NA" and "Unknown" metadata values separately on the Search page metadata sidebar.
- Moving a BED file filter widget on the pre-processing no longer causes the uploaded BED file to disappear.
- The "Get CSV" button on the Analysis Set details page no longer provides samples which the user has excluded from the analysis set.
- Eliminated a situation where sample metadata labels could obscure data points on the job results plot.