BokehJS 2.3.0 successfully loaded.

RNA-seq-like Gene Centric Signature Reverse Search (RGCSRS)

Input gene: USP28

USP28 is not a landmark gene.

USP28 is not an originally inferred L1000 gene.

USP28 is a newly inferred (our model) gene.

More information about USP28 can be found at the Gene and Drug Landing Page Aggregator

Input cell line: All

This Appyter provides visualizations of the top 5% of RNA-seq-like signatures induced by CRISPR knockouts and chemical perturbagens. Signatures are computed from transformed data profiles from the LINCS L1000 data. The transformation was performed using a two-step model:

  1. A cycleGAN model was used to first predict the RNA-seq expression of the 978 L1000 landmark genes
  2. A fully connected neural network was used to extrapolate the predicted RNA-seq expression of the 978 landmark genes to a full set of 23,164 genes

Signatures were computed using the characteristic direction method (Clark et al., 2014), as implemented here.

Each gene was pre-queried across all available RNA-seq-like signatures, and the top signatures where a gene is up or down-regulated are returned for each gene.

CRISPR KO signatures

Volcano Plots

In the following volcano plot, each point represents a single CRISPR knockout signature. The x-position indicates the log2(fold change) of the expression of the chosen gene in the signature, while the y-position indicates the absolute value of the characteristic direction coefficient of the chosen gene.

Note that the fold change and characteristic direction coefficients of the gene are not necessarily in the same direction for each signature; this is because in cases where a gene is both up- and down-regulated between replicate samples, the characteristic direction method prioritizes the more consistent direction of movement, which may not be consistent with the fold change. To read more about the characteristic direction method, please refer to Clark et al., 2014.

Points with same-direction fold change and CD coefficient values are highlighted by coloring them blue (up-regulated) or red (down-regulated). Darker colored points indicate higher differential expression of the gene in the corresponding signature.

Drag the plot to pan around. Use the toolbar to the right of the plot to zoom, reset the plot view, or download the plot.

Tables

The tables below display the characteristic direction (CD) coefficients, fold change values, and log2(fold change) values correponding to the expression of the chosen gene in each CRIPSR KO signature.

The rank of the gene in the signature is determined by its fold change relative to the fold change of the other genes that are regulated in the same direction; if a gene is ranked 1 in a signature where the gene is up-regulated, that means that out of all genes up-regulated in the signature, the input gene had the highest fold change and was the most up-regulated.

While only the top 10 signatures for each direction are displayed, below each table is a link to download the top 50 signatures for each direction.

A link to the Enrichr analysis results of the top 20 unique perturbations from the top signatures that up or down-regulate the input gene can be found below each table as well.

Top CRISPR KO signatures where USP28 is up-regulated (based on fold change)
CD Coefficient Fold Change Log2(Fold Change) Rank in Signature KO Gene Cell Line Timepoint
Signature
XPR040_U251MG.311_96H_G24_ASAH1 0.0198 1.1377 0.186066 5497.0 ASAH1 U251MG.311 96h
XPR028_A375.311_96H_I19_CYBB 0.0216 1.0977 0.134487 3897.0 CYBB A375.311 96h
XPR013_A375.311_96H_N19_CCL5 0.0200 1.0825 0.114382 3466.0 CCL5 A375.311 96h
XPR016_BICR6.311_96H_H15_EPHA1 0.0227 1.0727 0.101222 2596.0 EPHA1 BICR6.311 96h
XPRJJ001_A375_96H_G08_KDM5B 0.0222 1.0702 0.097895 2550.0 KDM5B A375 96h
XPR025_AGS.311_96H_B18_TET2 0.0257 1.0611 0.085575 2974.0 TET2 AGS.311 96h
XPR034_U251MG.311_96H_I13_TLK1 0.0219 1.0609 0.085291 6667.0 TLK1 U251MG.311 96h
XPR035_MCF7.311_96H_M06_PDXK 0.0201 1.0574 0.080479 3065.0 PDXK MCF7.311 96h
XPR018_HT29.311_96H_N16_JAML 0.0204 1.0553 0.077625 3900.0 JAML HT29.311 96h
XPR042_A549.311_96H_N11_OXCT1 0.0268 1.0544 0.076476 5208.0 OXCT1 A549.311 96h
Top CRISPR KO signatures where USP28 is down-regulated (based on fold change)
CD Coefficient Fold Change Log2(Fold Change) Rank in Signature KO Gene Cell Line Timepoint
Signature
XPR023_MCF7.311_96H_G24_RSL1D1 -0.0234 0.8967 -0.157262 3221.0 RSL1D1 MCF7.311 96h
XPR040_HS944T.311_96H_C02_ATP5B -0.0221 0.9234 -0.115000 3668.0 ATP5B HS944T.311 96h
XPR025_A549.311_96H_O22_STX4 -0.0232 0.9259 -0.111122 2024.0 STX4 A549.311 96h
XPR020_HT29.311_96H_O23_MANSC1 -0.0203 0.9375 -0.093076 3432.0 MANSC1 HT29.311 96h
XPR037_PC3.311B_96H_P02_GNAI3 -0.0198 0.9465 -0.079369 3925.0 GNAI3 PC3.311B 96h
XPR034_A549.311_96H_A03_COASY 0.0204 0.9468 -0.078895 5737.0 COASY A549.311 96h
XPR030_A375.311_96H_K15_GJB5 -0.0196 0.9485 -0.076311 3675.0 GJB5 A375.311 96h
XPR010_PC3.311B_96H_P21_PDHB -0.0201 0.9507 -0.072922 3521.0 PDHB PC3.311B 96h
XPR026_A549.311_96H_B16_UHRF1BP1 -0.0200 0.9509 -0.072563 3383.0 UHRF1BP1 A549.311 96h
XPR019_U251MG.311_96H_K15_KLF10 -0.0240 0.9530 -0.069431 4069.0 KLF10 U251MG.311 96h
---------------------------------------------------------------------------
Exception                                 Traceback (most recent call last)
Input In [14], in <cell line: 5>()
      3 display(HTML(down_xpr[:10].to_html(escape=False, col_space=70)))
      4 display(HTML(download_link(down_xpr[:100], f"{gene}_DnReg_L1000_CRISPR_signatures.tsv")))
----> 5 display(HTML(enrichr_link('CRISPR', xpr_gene_data, 'down', gene)))

Input In [11], in enrichr_link(pert, df, direction, gene)
     30     top_perts = comb_df.sort_values(by='FC', ascending=True) \
     31         .drop_duplicates(subset=['pert'],keep='first')['pert'][:20]
     32 pert_type = 'CRISPR target genes' if pert == 'CRISPR' else 'chemical compounds'
---> 33 results_url = enrichr(pert, top_perts, direction)
     34 return f'<a href={results_url} target="_blank">Enrichr analysis of top 20 {pert_type} that {direction}-regulate {gene}</a>'

Input In [11], in enrichr(pert, top_perts, direction)
     14 response = requests.post(list_url, files=payload)
     15 if not response.ok:
---> 16     raise Exception('Error analyzing gene list')
     17 time.sleep(0.5)
     18 return f"{enrich_url}?dataset={response.json()['shortId']}"

Exception: Error analyzing gene list

Chemical perturbation signatures

Volcano Plots

In the following volcano plot, each point represents a single chemical perturbation signature. The x-position indicates the log2(fold change) of the expression of the chosen gene in the signature, while the y-position indicates the absolute value of the characteristic direction coefficient of the chosen gene.

Note that the fold change and characteristic direction coefficients of the gene are not necessarily in the same direction for each signature; this is because in cases where a gene is both up- and down-regulated between replicate samples, the characteristic direction method prioritizes the more consistent direction of movement, which may not be consistent with the fold change. To read more about the characteristic direction method, please refer to Clark et al., 2014.

Points with same-direction fold change and CD coefficient values are highlighted by coloring them blue (up-regulated) or red (down-regulated). Darker colored points indicate higher differential expression of the gene in the corresponding signature.

Drag the plot to pan around. Use the toolbar to the right of the plot to zoom, reset the plot view, or download the plot.

Tables

The tables below display the characteristic direction (CD) coefficients, fold change values, and log2(fold change) values correponding to the expression of the chosen gene in each chemical perturbation signature.

The rank of the gene in the signature is determined by its fold change relative to the fold change of the other genes that are regulated in the same direction; if a gene is ranked 1 in a signature where the gene is up-regulated, that means that out of all genes up-regulated in the signature, the input gene had the highest fold change and was the most up-regulated.

While only the top 10 signatures for each direction are displayed, below each table is a link to download the top 50 signatures for each direction.

A link to the Enrichr analysis results of the top 20 unique perturbations from the top signatures that up or down-regulate the input gene can be found below each table as well.