Interactive Tools
There are several no-code options for exploring JUMP data. These are very useful for querying perturbations without needing experience in data analysis or programming. Detailed usage instructions are below this brief overview listing:
JUMP Round-Robin (jump_rr) tools
- Type in your gene/compound of interest and retrieve a list of similar genes/compounds
- Tools to browse images and inspect enriched features
- Visualize images of the perturbed cells
- No account creation required
Reference tools If you are not sure if your genes or compounds of interest are in JUMP, you can check it usingn one of our tools embedded on this website.
- Gene availability: Is my gene in a JUMP (or JUMP-adjacent) dataset?
- Compound availability: Is my compound in JUMP? If not, what is the closest equivalent?
Morpheus
- Explore matrices of data and perform basic calculations
- Explore heatmaps showing similarity among clustered samples
- No account creation required
Ardigen phenAID JUMP-CP Explorer
- Search similarities among perturbations and look at their images
- Account creation required, but access is free
Details and usage
JUMP_rr tools
A collection of online tables to explore, visualize and compare perturbations and their effect in cells.
Perturbation availability
Gene availability
A table to query which genes are available in the JUMP datasets. It contains the JUMP dataset alongside others with data from genetic perturbations.
- Link to the gene availability table
- Tutorial and details
Compound availability
A query tool to find the JUMP id or its most similar equivalent using an InChiKey or common name for a chemical.
Morpheus
- How can I use Morpheus to view relationships among samples?
Download the ORF and/or CRISPR data from https://doi.org/10.5281/zenodo.14025602 or from https://zenodo.org/records/14165010 (a smaller file with only genes having cosine similarity > 0.5 with other genes).
Drag and drop one of those files into Morpheus (at https://software.broadinstitute.org/morpheus/ in your web browser; no need to install it)
Go to Tools > Similarity Matrix and choose Metric = cosine similarity and Compute matrix for = Columns
Cluster the genes using Tools > Hierarchical clustering (we recommend 1 minus Pearson correlation, but other options are fine) for both rows and columns
Click the gear icon towards the top right and choose “Annotations” tab
Choose Symbol (for Rows and Columns) so that gene names will appear. You can also choose various annotations of genes, like biological process or molecular function but those names are long and make it hard to see unless you have a huge screen.
Play around!
- To find a gene of interest, in the search box type Symbol: then start typing the name and it will appear as highlighted in the scroll bar so you can scroll to find it. But if you are looking for top matches to a gene, note that clustering can have weird behavior, putting the closest match pretty far away depending on the other data points. The broad.io Datasette links (JUMPer tools) more directly answer the “top/bottom matches” question and take less hands-on time to configure and less compute time to load.
- doubleclick a gene name and the data matrix will sort to show you the closest matches on one end and the strongest anti-correlators at the other. To go back to your original view, just keep clicking the same gene you originally sorted by, it will change back to the previous sort eventually.
- note that in clustered view you can’t see anti-correlations which anecdotally seem more strong/common in CRISPR data. So clusters are a bit more interesting in ORF data.
- If you would like to trim the similarity matrix to only show genes with a strong match, you can follow this procedure:
- Add a new annotation by going to Tools > Create Calculated Annotation, with the name ‘#Matches’ with the formula COUNTIF(“>0.7”) This counts the number of genes where the similarity is above 0.7 (you can pick whatever threshold you prefer). Repeat this step for both rows and columns. Warning: although this seems like a simple step, for some reason it is fairly intensive and might require a suitably powerful computer - if the browser asks to Wait or Exit, waiting sometimes works.
- Filter samples where ‘#Matches’ >=2 (because every sample matches itself); again repeat this for both rows and columns.
- This will yield ~350 genes in the ORF data, for example. You can increase the number of matches to filter (to 3 or 4, as you prefer).
- Add a new annotation by going to Tools > Create Calculated Annotation, with the name ‘#Matches’ with the formula COUNTIF(“>0.7”) This counts the number of genes where the similarity is above 0.7 (you can pick whatever threshold you prefer). Repeat this step for both rows and columns. Warning: although this seems like a simple step, for some reason it is fairly intensive and might require a suitably powerful computer - if the browser asks to Wait or Exit, waiting sometimes works.
- doubleclick a gene name and the data matrix will sort to show you the closest matches on one end and the strongest anti-correlators at the other. To go back to your original view, just keep clicking the same gene you originally sorted by, it will change back to the previous sort eventually.
- To find a gene of interest, in the search box type Symbol: then start typing the name and it will appear as highlighted in the scroll bar so you can scroll to find it. But if you are looking for top matches to a gene, note that clustering can have weird behavior, putting the closest match pretty far away depending on the other data points. The broad.io Datasette links (JUMPer tools) more directly answer the “top/bottom matches” question and take less hands-on time to configure and less compute time to load.