Goal:
- Characterization of selective Protein kinase inhibitors using Cellpainting assay
- Identifying novel phenotypes associated with each kinase inhibitor
Welch ttest used to perform pairwise feature comparison
Total no of features = 595
1. DMSO and BSJ-03-136
2. DMSO and BSJ-04-030
95th percentile threshold of Control_Active {-log10[p-values]} used to select features
Features list:
['Cytoplasm_Correlation_Correlation_DNA_Mito',
'Cytoplasm_Correlation_Correlation_DNA_RNA',
'Cells_Intensity_MedianIntensity_DNA',
'Cells_Intensity_MADIntensity_DNA',
'Cells_Intensity_MinIntensityEdge_RNA',
'Cells_Texture_Gabor_RNA_20',
'Cells_Texture_InverseDifferenceMoment_RNA_5_0',
'Cells_Texture_Gabor_Mito_20',
'Cells_Texture_Gabor_ER_20',
'Cells_Correlation_Correlation_DNA_Mito',
'Cells_Intensity_MeanIntensity_ER',
'Cytoplasm_Texture_Gabor_AGP_5',
'Cytoplasm_Intensity_UpperQuartileIntensity_DNA',
'Cells_Texture_Contrast_DNA_5_0',
'Cells_Correlation_Correlation_DNA_ER',
'Cells_Correlation_Correlation_DNA_RNA',
'Cytoplasm_Texture_DifferenceVariance_AGP_20_0',
'Cytoplasm_Texture_Contrast_Mito_10_0',
'Cells_Texture_Gabor_RNA_10',
'Cells_Texture_DifferenceVariance_Mito_20_0',
'Cells_Intensity_UpperQuartileIntensity_DNA',
'Nuclei_Neighbors_SecondClosestObjectNumber_2',
'Cells_Texture_Gabor_AGP_20',
'Cells_Texture_Correlation_RNA_10_0',
'Cytoplasm_AreaShape_Solidity',
'Cytoplasm_Intensity_MADIntensity_DNA',
'Cells_Texture_Correlation_RNA_5_0',
'Cells_Texture_Gabor_RNA_5',
'Cells_Texture_DifferenceVariance_Mito_10_0',
'Cytoplasm_Texture_Gabor_RNA_5']
The main aim of this project to identify protein kinase inhibitors (Active, Inactive forms) which produce distinct cellpainting profiles in two cancer cell lines (A549, U2OS)
Drugs(Active, Inactive) forms targeting DCLK1 showed similar profiles in both cell lines
AX15836 (active inhibitor of ERK5) didn't work and clustered with DMSO samples
while JWG-071 (active) and JWG-119 (inactive) showed slight difference in profiles in both cell lines
We got only one compound of FAK inhibitor
Both active and inactive compounds targeting PIN1 clustered with DMSO
FMF-04-159-2(covalent) and FMF-05-176-1 (reversible) drugs targeting CDK14pan-TAIRE displayed distinct profiles only in A549 cells while FMF-05-176-1 (reversible) in U2oS cells didn't seem to work
active and inactive Protein kinases targeting SECRET pathway displayed distinct profiles only in A549 cells
Features are grouped together for AreaShape and Neighors category and absolute difference of mean of z-scores values are calculated for:
DMSO Controls
FMF-04-159-2 Covalent
FMF-05-176-1 Reversible
AreaShape & Neighbors Features
All other Features
Top 10 Intensity features in comparison of Covalent vs Reversible drug
Top 10 RadialDistribution Features of Mito Channel
Top 10 RadialDistribution Features of RNA Channel
**Top 6 Nuclei AreaShape Features **
Features are grouped together for AreaShape and Neighors category and absolute difference of mean of z-scores values are calculated for:
Note: we do observe differences in Cell and Cytoplasm AreaShape features more obvious in Active Protein Kinase inhibitor (PK) compared with the inactive form of the inhibitor. There is no such obvious differences observed between active and inactive PK inhibitors. Neighbors features also have high values??? suggesting more density of cells?
Other Feature groups
1. Overall differences are higher in Active PK inhibitor relative to Controls than Inactive vs controls
2. Intensity features of DNA, Mito, ER and Radial features of Mito, ER and RNA showed high differences relative to controls
3. Granularity features for all channels showing differences relative to controls
4. However, Intensity and Radia ldistribution features of Mito and ER showed high differences in Active vs Inactive Protein kinase inhibitor.
Features
Category:"Neighbors"
list of differential features between active and inactive compound of SECRET pathway
['Cells_Neighbors_FirstClosestDistance_Adjacent',
'Nuclei_Neighbors_SecondClosestObjectNumber_2',
'Nuclei_Neighbors_AngleBetweenNeighbors_2',
'Cells_Neighbors_AngleBetweenNeighbors_Adjacent',
'Cells_Neighbors_PercentTouching_Adjacent']
Category: "RadialDistribution" Channel: "Mito"
Category: "RadialDistribution" Channel: "ER"
Category: "Intensity" Channel: "Mito"
Category: "Intensity" Channel: "ER"
Goals: Exploratory analysis to ensure everything is working as expected and look for relationships between perturbations
Conclusion:
Two independent unsupervised clustering approaches revealed that following two Protein Kinase inhibitors targeting (SECRET pathway) have distinct morphological profiles from DMSO as well as from each other in A549 cell line. However this is not seen in case of U2oS cell line.
Note: Single Cell clustering revealed interesting pattern in Active compound most of the cell profiles are pushed on the top of UMAP space. While in case of DMSO few no of cell occupy the same space. In case of inactive compound it has showed mixed response.
Plates: 2
BR00100032
BR00100037
s3://imaging-platform/projects/2018_11_20_GeneCpdFollowup/workspace/backend/2018_11_20_Batch1/
Platemap:
Both plates have same platemap
Celllines:
Positive Controls:
C3 -------- BRD-K97963946-001-01-3 [ Drug Repurposing Hub]
C4 -------- BRD-K97309399-001-09-4 [Drug Repurposing Hub
Negative Controls:
C1 -------- BRD-A15435692-003-02-3 [Drug Repurposing Hub]
C2 -------- BRD-K90789829-001-07-8 [ Drug Repurposing Hub]
Technical Replicates per plate:
DMSO: 272
Positive controls : 4
Tested Drugs : 8
Features are grouped together for AreaShape and Neighors category and absolute difference of mean of z-scores values are calculated for:
There were two active drugs AX15836 and JWG-071 and one inactive JWG-119 targeting ERK5 kinases. However,
one active AX15836 drug didn't seems to work in either of the cell lines
In creating feature grid, we are using only
**JWG-071 ** Active
JWG-119. Inactive
Cytominer workflow used for preprocessing, Feature normalization and selection and mean aggregated well profiles were created
Note: As in this experiment we randomly sampled 112 DMSO's from each plate and used them for normalization of variables
A total of 598 Features selected and used in downstream analysis
Median replicate correlation analysis is performed and most of compounds have high replicate correlative with the exception of two compounds both of them displayed low replicate correlation in both celllines
Similarity correlation analysis is performed where each row represents each replicate well sample and each column represent features. Protein kinase inhibitors, in the top and bottom highlighted region displayed distinct profile compared to DMSO, shown in the middle of the figure.
Similarity correlation analysis for aggregated profiles for each compound
A549
U2oS
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.