This site is best supported with Google Chrome/Chromium.

How to use this tool?

This tool was originally designed by Chen et al. (2023) with the goal of streamlining the selection of marker gene combinations by using a Bayesian mixture model to binarize gene expression.

We recommend users start from a cluster of interest (From cluster) or from a gene that marks several clusters of interest (From gene). Either tab will help users identify off-target clusters that could be marked when using only one marker and help to select additional markers to uniquely and specifically label cluster(s) of interest.

Once marker combinations are found, users might want to see how other clusters in the same dataset express this combination (Co-expression plot).

While binarization greatly reduces computational complexity, it has its limitations: because gene expression is a spectrum, it can be difficult to categorize expression as either ON or OFF in some clusters. We thus encourage users to also examine log-normalized expression value and its dynamic (Plot expression trend) and decide whether a satisfactory pair of markers has been identified.

Shared options

Which dataset to use?

This tool includes two separate datasets, one including dsx-expressing neurons of males of four Drosophila species (D. melanogaster, D. yakuba, D. santomea, and D. teissieri) and one including dsx-expressing neurons of D. melanogaster males and females. Users are encouraged to select the dataset they are most interested in.

Clusters of interest

The clusters to focus on (Find distinct genes within) or to highlight (Plot expression trend).

Probability threshold to be considered as expressed

Binomial probability of a gene being expressed in a cluster/cell type. A cutoff is selected here to decide the probability threshold to consider a gene as ON.

From cluster

Select how many genes to show and which species or sex to include, depending on the dataset. After clicking Show genes, the consistently expressed genes will be shown and ranked by how many other clusters they are also detected in.

You can select a gene and move forward to Find distinct genes within.

From gene

Similar to From cluster but starting from a gene.

Users can select which species or sex to include, depending on the dataset. After clicking Show clusters, all clusters that consistently express this gene will be shown, and users can choose the clusters to highlight and move forward to Find distinct genes within.

Find distinct genes within

Among clusters of interest, find genes that are only ON in one cluster but not others. Use the check boxes to select the species or sex a gene must be expressed in to be considered as ON.

Click Find genes to visualize how well these clusters can be distinguished with the addition of another gene. Users could pick a gene from this list to use together with which genes were used in From cluster or From gene.

Users might want to visualize gene pairs to see how they express in other clusters in the dataset with Co-expression plot.

Co-expression plot

Select genes of interest and visualize their co-expression pattern.

Plot expression trend

This tool generates an interactive line plot of log-transformed cluster average normalized expression. Pointing the cursor at a point will show the cluster labels. Clusters of interest will be highlighted (and coded with distinct colors when you download a static version of the plot).