Comments (9)
Hi yong114,
Thank you for using celloracle!
Could you please tell me about your computational environment?
I want to know these things below.
- Operating system name
- OS version.
- CellOracle version
- Python version,
- Jupyter notebook version
from celloracle.
Hi Kenji,
Thank you so much for your prompt reply! The pipeline was running smoothly until the network calculation. I installed the CellOracle on the cluster in the Conda environment.
Please find the information as below as requested:
Operating system name - CentOS Linux release 7.6.1810 (Core)
CellOracle version - 0.3.1
Python version - 3.6.10
Jupyter notebook version - 1.0.0
Sorry I have one more question, I have log-transformed and scaled my data as in tutorial. Would that be the problem or my datasets too noisy please?
Thank you very much for your help again!
from celloracle.
Thank you for letting me know about your environment.
I'm still not 100% sure why you got that error, I tried to update our code to solve the problem.
Could you please re-install celloracle and try the analysis again with the new version of celloracle (0.3.7)?
from celloracle.
Let me answer your second question about log transformation and scaling.
Although we did log-transformation and scaling in the tutorial, these processes were for the calculations of dimensional reduction and clustering, and these were not for the celloracle GRN calculation.
We need non-transformed data for Network analysis.
And we recover non-transformed data just before GRN calculation in our tutorial notebook.
If you are doing the same thing as this tutorial notebook, you have already used non-transformed and non-scaled data for network analysis.
If you skipped this process, please do that.
from celloracle.
Thank you so much for your kind response! Much appreciated!
I will try to work on that with the latest version, thanks for the explanation.
Will update more soon once I run the analysis.
Thank you again.
from celloracle.
Hi Kenji,
Many thanks again for your responses. The latest version CellOracle (0.3.7) solved the problem and I have successfully running the analysis. I'm sorry, I have a few questions regarding to the interpretation of the final results please with the Sankey diagram in your tutorial.
Based on the tutorial, I was wondering if GATA1 also involved in the cell transition between Megakaryoctes and GMP (pink flow) / and also between Granulocytes and late GMP (red flow) please (based on the colour flow charts?) Sorry for my questions.
Thank you very much again.
from celloracle.
Hi Yong,
I'm glad to hear that CellOracle worked in your environment!
The answer to your question is Yes.
Maybe the Gata1 KO simulation suggests that Gata1 KO can induce cell transition from MK to GMP and also Gata1 KO can induce cell state transition from Granulocyts to GMP.
I think these predictions are convincing because it seems that there are several experimental studies to support these predictions.
https://pubmed.ncbi.nlm.nih.gov/11012179/
https://www.frontiersin.org/articles/10.3389/fmed.2017.00115/full
FYI, please let me add some tips below. These tips might help you interpret the results of cell oracle's simulation.
-
[Tips to interpret sankey diagram] The snaky diagram is a summary of Markov simulation. In this simulation, we simulate cell transition when we perturb a TF(s). We calculate transition probably using GRN, and we use scRNA-seq data for the initiation state.
The left side of the sankey diagram is a cell state at the initiation, which is intact scRNA-seq data. The right side of the sankey diagram is a predicted final cell state after Malkov simulation.
If you see a transition from Mk to GMP, it means Mk might become like a GMP if we perturb TF. But, the sankey diagram only shows information about initial state and terminal state, it does not always means cells directly convert from cell MK to GMP. There is a possibility that the cells change through a different cell type before reaching the final state. For example, Mk might change into MEP, then this MEP change into GMP. Again, please keep in mind that the sankey diagram only shows information about initial state and terminal state. And the results of this simulation depends on the number of steps in the Markov simulation. Please star the Markov simulation with a small number of steps. -
[Difference between cell oracle's simulation and experimental perturbation] You might want to predict the phenotype of TF KO. But please keep in mind that CellOracle's simulation is not exact same as experimental KO in some viewpoints.
First, CellOracle's simulation uses scRNA-seq data for the initiation state, but your initial state of KO might be different from the scRNA-seq data. In the simulation with the hematopoiesis dataset, we did the Gata1 KO simulation. In the simulation, the Gata1 gene was deleted "in the middle of " hematopoiesis. However, when you perform mouse KO experiment with Gata1 KO mouse, gene KO was "already done at the beginning" of hematopoiesis. Thus, the condition of the simulation is not always same as actual experiments.
We designed celloracle to understand the mechanism how GRN control cell identity intuitively. The aim of this project is not to predict the exact same phenotype of perturbation experiments. Thus, please use celloracle to analyze the GRN and cell identity in your scRNA-seq data rather than to predict the exact same phenotype of perturbation experiments. -
[Potential limitation of celloracle] The simulation is performed based on GRN connectivity. If you try simulation with a TF that have small number of connection, the accuracy could be bad. Please make sure that your TF of interest have large network degree before doing simulation. Also, celloracle cannot predict the phenotype of a TF if the TF's regulatory mechanism is strongly affected by the post-transcriptional regulation. It is the limitation of scRNA-seq experiments itself.
-
[Resource] We provide information about CellOracle in various format. I hope these resources help your analysis.
YouTube channel: https://www.youtube.com/channel/UCVy_6MUYL594cyNijzD0EVg
Web documentation: https://morris-lab.github.io/CellOracle.documentation/index.html
Preprint: https://www.biorxiv.org/content/10.1101/2020.02.17.947416v3
from celloracle.
Hi yong114,
Yes. You're right.
For the simulation, TF need to have decent number of connections in at least one cluster.
If your TF of interest has no connections in a cluster, there are several possible interpretations.
- The TF does not have significant biological role in the cluster.
- CellOracle could not detect connections between the TF and its target gene if the genes are dominantly regulated by a post-transcriptional mechanism. CellOracle was designed to analyze only transcriptional GRN.
- The connection for a TF cannot be detected if base GRN (the output of scATAC-seq analysis) does not contain any data for the TF.
- Technical error.
from celloracle.
Thank you so much Kenji for your kind explanation! It was really helpful for me!
Thank you again for your time. Cheers.
from celloracle.
Related Issues (20)
- Request for Schistosoma japonicum genome
- WinError: FileNotFound at installation HOT 1
- tfinfo and GRN files HOT 4
- Plot cell types under 'development' flow grid
- Problem when using a custom genomes_dir for TF scan HOT 2
- Request for 3D UMAP
- model overfitting
- Heterogeneous dataset subsetting and number of cells/cluster
- Loading CellOracle takes too much time.
- Loading CellOracle takes too much time
- We don't have default motifs for your species, Please specify motif data by yourself. HOT 1
- Can't convert from Seurat to anndata #8928
- GimmeMotifs installed but can't install celloracle because of 'Failed building wheel for gimmemotifs' HOT 1
- Saving plots as individual files?
- Does celloracle need pseudotime information to give cluster wise GRNs?
- does not have enough regulatory connection in the GRNs. Cannot perform simulation.
- Error in calculating embedding shift HOT 3
- The Kernel crashed when I run tfi.scan
- seuratToAnndata
- Producing a video visualisation (like shown on https://celloracle.org/)
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from celloracle.