Start wilms analysis #681
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,8 @@ | ||
| # Ignore everything by default | ||
| * | ||
|
|
||
| # Include specific files in the docker environment | ||
| !/renv.lock | ||
| !/requirements.txt | ||
| !/environment.yml | ||
| !/conda-lock.yml |
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,7 @@ | ||
| # Results should not be committed | ||
| /results/* | ||
| !/results/README.md | ||
|
|
||
| # Ignore the scratch directory (but keep it present) | ||
| /scratch/* | ||
| !/scratch/.gitkeep |
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,19 @@ | ||
| # pull base image | ||
| FROM bioconductor/tidyverse:3.19 | ||
|
There was a problem hiding this comment. I had a problem building this locally, but it seems that the base image might be the issue. I will need to investigate a little more. There was a problem hiding this comment. I changed for this: # Base image on the Bioconductor 3.19 image
FROM bioconductor/r-ver:3.19and build locally like this podman buildx build . -t openscpca/cell-type-wilms-tumor-06:latest --platform linux/amd64and it works :) There was a problem hiding this comment. I am building locally to make some tested recommendations regarding what to do with There was a problem hiding this comment. Thank you that would be great! I am initiating a renv environment from the current R Session to then simplify the Dockerfile using the renv.lock (https://openscpca.readthedocs.io/en/latest/ensuring-repro/docker/docker-images/) renv_init() is taking a while... I'll continue tomorrow :) There was a problem hiding this comment. I had trouble accessing RStudio on the image I built locally. I'm attempting to isolate the problem... but that unfortunately requires rebuilding things 😅 ⏳ There was a problem hiding this comment. Thank you for being on it :) I have troubles trying to implement the renv() environment and to change for the bioconductor/r-ver:3.19 based image... So far I can only open RStudio using the following Dockerfile (just a bit "cleaned" compared to the PR). # Base image on the Bioconductor 3.19 image
FROM bioconductor/tidyverse:3.19
# Set global R options
RUN echo "options(repos = 'https://cloud.r-project.org')" > $(R --no-echo --no-save -e "cat(Sys.getenv('R_HOME'))")/etc/Rprofile.site
ENV RETICULATE_MINICONDA_ENABLED=FALSE
RUN R --no-echo --no-restore --no-save -e "install.packages('remotes')"
RUN R -e "devtools::install_github('enblacar/SCpubr')"
RUN R -e "remotes::install_github('satijalab/seurat', 'seurat5', quiet = TRUE)" # this also install patchwork (and others)
RUN R -e "remotes::install_github('satijalab/azimuth', quiet = TRUE)" # this also install SingleCellExperiment, DT (and others)
RUN R -e "remotes::install_github('cancerbits/DElegate')"
RUN R -e "install.packages('viridis')"
RUN R -e "install.packages('ggplotify')"
RUN R -e "BiocManager::install('edgeR')"
# make sure all R related binaries are in PATH in case we want to call them directly
ENV PATH ${R_HOME}/bin:$PATHThere was a problem hiding this comment. @jaclyn-taroni FYI, I removed from the run.sh et config file unnecessary volume and path to data that are specific to our group. I realized it prevent the execution of the docker image if not defined! Just in case it can help. There was a problem hiding this comment. Last comment of the week ;) I added the renv.lock file from the RStudio session of docker image I am running on my machine. I am now trying to use it to build the image as described here: (https://openscpca.readthedocs.io/en/latest/ensuring-repro/docker/docker-images/), but building of the image is full of ERROR because of BiocManager version not matching the bioconductor version.. (as described here rstudio/renv#517) Will take some time of fine tunning ⏳ But I think it's worth trying to have the Docker image in this format, then it might be easier to share/reproduce? What do you think @jaclyn-taroni ? There was a problem hiding this comment. I agree that our end goal should be to use I am specifically struggling with what base image should be used right now. I expect folks working on this project (including all of our staff at the Data Lab) to be on Macs with Apple Silicon. My understanding is that you want to be able to develop using the RStudio Server IDE from within the running container. This is often part of my workflow, and I expect many other project participants might have this use case! (That is to say, I think we are bumping into a problem that will come up again and again in the project...) However, I'm not sure we can use So, we might instead want to use the And until we implement Then I believe we'd have an image we can build and push to Elastic Container Registry using GitHub Actions that can also be built and used locally on ARM machines. This compatibility seems like a good goal to me. I would have liked to test this all conclusively, but installing Azimuth is taking a very long time for me 😅 Appreciate your patience, @maud-p! |
||
|
|
||
| # Set global R options | ||
| RUN echo "options(repos = 'https://cloud.r-project.org')" > $(R --no-echo --no-save -e "cat(Sys.getenv('R_HOME'))")/etc/Rprofile.site | ||
| ENV RETICULATE_MINICONDA_ENABLED=FALSE | ||
|
|
||
| RUN R --no-echo --no-restore --no-save -e "install.packages('remotes')" | ||
|
|
||
| RUN R -e "devtools::install_github('enblacar/SCpubr')" | ||
| RUN R -e "remotes::install_github('satijalab/seurat', 'seurat5', quiet = TRUE)" # this also install patchwork (and others) | ||
| RUN R -e "remotes::install_github('satijalab/azimuth', quiet = TRUE)" # this also install SingleCellExperiment, DT (and others) | ||
| RUN R -e "remotes::install_github('cancerbits/DElegate')" | ||
| RUN R -e "install.packages('viridis')" | ||
| RUN R -e "install.packages('ggplotify')" | ||
| RUN R -e "BiocManager::install('edgeR')" | ||
|
|
||
| # make sure all R related binaries are in PATH in case we want to call them directly | ||
| ENV PATH ${R_HOME}/bin:$PATH | ||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,149 @@ | ||
| # Wilms Tumor Dataset Annotation (SCPCP000006) | ||
|
|
||
| Wilms tumor (WT) is the most common pediatric kidney cancer characterized by an exacerbated intra- and inter- tumor heterogeneity. | ||
| The genetic landscape of WT is very diverse in each of the histological contingents. | ||
| The COG classifies WT patients into two groups: the favorable histology and diffuse anaplasia. | ||
| Each of these groups is composed of the blastemal, epithelial, and stromal populations of cancer cells in different proportions, as well as cells from the normal kidney, mostly kidney epithelial cells, endothelial cells, immune cells and normal stromal cells (fibroblast). | ||
|
|
||
| ## Description | ||
|
|
||
| Here, we first aim to annotate the Wilms Tumor snRNA-seq samples in the SCPCP000006 (n=40) dataset. To do so we will: | ||
|
|
||
| • Provide annotations of normal cells composing the kidney, including normal kidney epithelium, endothelium, stroma and immune cells | ||
| • Provide annotations of tumor cell populations that may be present in the WT samples, including blastemal, epithelial, and stromal populations of cancer cells | ||
| Based on the provided annotation, we would like to additionally provide a reference of marker genes for the three cancer cell populations, which is so far lacking for the WT community. | ||
|
|
||
|
|
||
| The analysis is/will be divided as the following: | ||
|
|
||
| - [x] Metadata file: compilation of a metadata file of marker genes for expected cell types that will be used for validation at a later step | ||
| - [ ] Script: clustering of cells across a set of parameters for few samples | ||
| - [ ] Script: label transfer from the fetal kidney atlas reference using runAzimuth | ||
| - [ ] Script: run InferCNV | ||
| - [ ] Notebook: explore results from steps 2 to 4 for about 5 to 10 samples | ||
| - [ ] Script: compile scripts 2 to 4 in a RMardown file with required adjustements and render it across all samples | ||
| - [ ] Notebook: explore results from step 6, integrate all samples together and annotate the dataset using (i) metadatafile, (ii) CNV information, (iii) label transfer information | ||
|
|
||
| ## Usage | ||
| From Rstudio, run the Rmd reports or render the R scripts (see below R studio session set up). | ||
| Please before running the script, make sure that the paths are correct. | ||
| You can also simply have a look at the html reports in the notebook folder. | ||
| Here, no need to run anything, we try to guide you through the analysis. Have a look at the code using the unhide code button on the top right of each chunk! | ||
|
|
||
| ## Input files | ||
|
|
||
| ### single nuclei data | ||
|
|
||
| We work with the _processed.rds SingleCellExperiment objects. | ||
| From the module directory, make sure that the conda environment is set-up: | ||
|
|
||
| ```shell | ||
| conda activate openscpca | ||
| ``` | ||
|
|
||
| log into AWS CLI: | ||
| ```shell | ||
| # replace `openscpca` with your AWS CLI profile name if it differs | ||
| export AWS_PROFILE=openscpca | ||
| aws sso login | ||
| ``` | ||
|
|
||
| use download-data.py to download the data as the following: | ||
| ```shell | ||
| ../../download-data.py --projects SCPCP000006 | ||
| ``` | ||
| This is saving the data in OpenScPCA-analysis/data/current/SCPCP000006 | ||
|
|
||
| Of note, this requires AWS CLI setup to run as intended: https://openscpca.readthedocs.io/en/latest/technical-setup/environment-setup/configure-aws-cli/ | ||
|
|
||
| ### sample metadata | ||
|
|
||
| The OpenScPCA-analysis/data/current/SCPCP000006/single_cell_metadata.tsv file contains clinical information related to the samples in the dataset. | ||
| Some information can be helpful for annotation and validation: | ||
|
|
||
| - treatment: Some of the samples have been pre-treated with chemotherapy and some are upfront resection. | ||
| We expect few changes between the 2 conditions, including a higher immune infiltration and more DNA damages pathways in treated samples. | ||
|
|
||
| - histology: the COG classifies Wilms tumor as either (i) Favorable or (ii) Anaplastic. | ||
| Some differenices are expected, some marker genes or pathways are associated with anaplasia (see sets of marker gene). | ||
|
|
||
| ## Output files | ||
|
|
||
| ## Marker sets | ||
|
|
||
| This folder is a resource for later validation of the annotated cell types. | ||
|
|
||
| ### The table CellType_metadata.csv contains the following column and information: | ||
| - "gene_symbol" contains the symbol of the described gene, using the HUGO Gene Nomenclature | ||
| - ENSEMBL_ID contains the stable identifier from the ENSEMBL database | ||
| - cell_class is either "malignant" for marker genes specific to malignant population, or "non-malignant" for markers genes specific to non-malignant tissue or "both" for marker genes that can be found in malignant as well as non-malignant tissue but are still informative in respect to the cell type. | ||
| - cell_type contains the list of the cell types that are attributed to the marker gene | ||
| - DOI contains the list of main publication identifiers supporting the choice of the marker gene | ||
| - comment can be empty or contains any additional information | ||
|
|
||
| |gene_symbol|ENSEMBL_ID|cell_class|cell_type|DOI|comment| | ||
| |---|---|---|---|---|---| | ||
| |WT1|ENSG00000184937|malignant|cancer_cell|10.1242/dev.153163|Tumor_suppressor_WT1_is_lost_in_some_WT_cells| | ||
|
There was a problem hiding this comment. I think of sc/snRNA-seq as better suited to picking up overexpression than loss. I know you've worked on these data before, so I'm just curious if you expect or have observed differences in WT1 expression in the cancer cells. Although, I see that you put:
in #635, so maybe we don't know yet! There was a problem hiding this comment. I added the (?) for the reason you mentionned, as we are looking for loss of function, I am not sure that we can really use it for annotation. However, about 20% of Wilms tumor would have imperment of WT1, and I would expect the WT1 mutated Wilms tumor to have a specific transcriptional program. At the final step of integration of the 40 samples together, I would expect a cluster negative for WT1. Also, the normal kidney should be WT1 positive. So in this last step I think looking at WT1 would make sense. |
||
| |IGF2|ENSG00000167244|malignant|cancer_cell|10.1038/ng1293-408|NA| | ||
| |TP53|ENSG00000141510|malignant|anaplastic|10.1158/1078-0432.CCR-16-0985|Might_also_be_in_small_non_anaplastic_subset| | ||
|
There was a problem hiding this comment. From the abstract of this publication, I wonder if looking at TP53 loss/activation at a pathway level would be interesting 🤔 There was a problem hiding this comment. This is a great idea, I will implement this in the next PR:
There was a problem hiding this comment. |
||
| |MYCN|ENSG00000134323|malignant|anaplastic|10.18632/oncotarget.3377|Also_in_non_anaplastic_poor_outcome| | ||
| |MAX|ENSG00000125952|malignant|anaplastic|10.1016/j.ccell.2015.01.002|Also_in_non_anaplastic_poor_outcome| | ||
| |SIX1|ENSG00000126778|malignant|blastema|10.1016/j.ccell.2015.01.002|NA| | ||
| |SIX2|ENSG00000170577|malignant|blastema|10.1016/j.ccell.2015.01.002|NA| | ||
|
Comment on lines
+89
to
+90
There was a problem hiding this comment. Similar to my TP53 comment – from a quick look at this publication, I wonder if looking at the altered expression patterns rather than the individual genes could be helpful. There was a problem hiding this comment. In the MSigdB C3/MIR gene sets, we have gene sets for DICER and DROSHA, I could also give a try to run enrichment for this dataset. There was a problem hiding this comment. |
||
| |CITED1|ENSG00000125931|malignant|blastema|10.1593/neo.07358|Also_in_embryonic_kidney| | ||
| |PTPRC|ENSG00000081237|immune|NA|10.1101/gr.273300.120|NA| | ||
| |CD68|ENSG00000129226|immune|myeloid|10.1186/1746-1596-7-12|NA| | ||
| |CD163|ENSG00000177575|immune|macrophage|10.1186/1746-1596-7-12|NA| | ||
| |VWF|ENSG00000110799|endothelium|endothelium|10.1134/S1990747819030140|NA| | ||
| |CD3E|ENSG00000198851|immune|T_cell|10.1101/gr.273300.120|NA| | ||
| |MS4A1|ENSG00000156738|immune|B_cell|10.1101/gr.273300.120|NA| | ||
| |FOXP3|ENSG00000049768|immune|T_cell|10.1101/gr.273300.120|Treg| | ||
| |CD4|ENSG00000010610|immune|T_cell|10.1101/gr.273300.120|NA| | ||
| |CD8A|ENSG00000153563|immune|T_cell|10.1101/gr.273300.120|NA| | ||
| |EPCAM|ENSG00000119888|NA|epithelial|10.1016/j.stemcr.2014.05.013|epithelial_malignant_and_non_malignant| | ||
| |NCAM1|ENSG00000149294|malignant|blastema|10.1016/j.stemcr.2014.05.013|might_also_be_expressed_in_non_malignant| | ||
| |PODXL|ENSG00000128567|non-malignant|podocyte|10.1016/j.stem.2019.06.009|NA| | ||
| |COL6A3|ENSG00000163359|malignant|mesenchymal|10.2147/OTT.S256654|might_also_be_expressed_in_non_malignant_stroma| | ||
| |THY1|ENSG00000154096|malignant|mesenchymal|10.1093/hmg/ddq042|might_also_be_expressed_in_non_malignant_stroma| | ||
|
There was a problem hiding this comment. I'm curious why this gene differs from some of the ones outlined in the abstract and is included. I suppose I would not expect the ones outlined in the abstract to be specific to malignant cells necessarily. There was a problem hiding this comment. unfortunately I don't know about one mesenchymal gene specific for mesenchymal Wilms tumor cells... For some colleagues who wanted to characterize CAF, the best approach I found was:
This is not perfect I think, but at least we should have clusters enriched in the target population. Stromal cells are really easily identified based on either few markers, or label transfer from the fetal kidney reference. They often form one single cluster. This is the reason why I didn't spend too much time adding marker genes for them, but I can add few more mesenchymal markers and references for correctness :) There was a problem hiding this comment.
There was a problem hiding this comment. I'd defer to you since you've spent more time thinking about this problem 😄 You don't need to add them – we'll have a record of this conversation! |
||
|
|
||
|
|
||
| ### The table GeneticAlterations_metadata.csv contains the following column and information: | ||
| - alteration contains the number and portion of the affected chromosome | ||
| - gain_loss contains the information regarding the gain or loss of the corresponding genetic alteration | ||
| - cell_class is "malignant" | ||
| - cell_type contains the list of the malignant cell types that are attributed to the marker gene, either blastemal, stromal, epithelial or NA if none of the three histology is more prone to the described genetic alteration | ||
| - DOI contains the list of main publication identifiers supporting the choice of the genetic alteration | ||
| - comment can be empty or contains any additional information | ||
|
|
||
| |alteration|gain_loss|cell_class|cell_type|DOI|PMID|comment | ||
| |---|---|---|---|---|---|---| | ||
| |11p13|loss|malignant|NA|10.1242/dev.153163|NA|NA| | ||
| |11p15|loss|malignant|NA|10.1128/mcb.9.4.1799|NA|NA| | ||
|
There was a problem hiding this comment. I get a 404 at: https://doi.org/10.1128/mcb.9.4.1799 – perhaps a typo in the DOI? There was a problem hiding this comment. oh sorry, complete doi is https://doi.org/10.1128/mcb.9.4.1799-1803.1989 There was a problem hiding this comment. should be corrected in the READ.ME and csv file :) |
||
| |16q|loss|malignant|NA|NA|1317258|Associated_with_relapse| | ||
| |1p|loss|malignant|NA|NA|8162576|Associated_with_relapse| | ||
| |1q|gain|malignant|NA|10.1016/S0002-9440(10)63982-X|NA|Associated_with_relapse| | ||
|
|
||
|
|
||
|
|
||
| ## Software requirements | ||
|
|
||
| To perform the analysis, run the RMarkdown script in R (version 4.4.1). | ||
| The main packages used are: | ||
| - Seurat version 5 | ||
| - Azimuth version 5 | ||
| - inferCNV | ||
| - SCpubr for visualization | ||
| - DT for table visualization | ||
| - DElegate for differential expression analysis | ||
|
|
||
| For complete reproducibility of the results, you can build and run the docker image using the Dockerfile. This will allow you to work on RStudio (R version 4.4.1) from the based image bioconductor/tidyverse:3.19. | ||
|
|
||
| In the config.yaml file, define your system specific parameter and paths (e.g. to the data). | ||
| Execute the run.sh file and open RStudio in your browser (http://localhost:8080/). | ||
| By default, username = rstudio, password = wordpass. | ||
|
|
||
|
|
||
|
There was a problem hiding this comment. Can you add an H4 section here on Halbritter lab internal development please? I'd expect that would include how to run the script (this is taking into account some review feedback): This hopefully helps you with your own development if, for example, you go on vacation for two weeks and come back to this! 😄 But it also helps others understand that this bash script isn't for their use. There was a problem hiding this comment. I'll have to ask for some help in the Halbritter lab, add it on my to do list ;) |
||
|
|
||
|
|
||
| ## Computational resources | ||
|
|
||
| No need to run any analysis, just open the metadata table! | ||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,12 @@ | ||
| project_name: maudp_ScPCA | ||
| project_docker: cancerbits/dockr:maudp_ScPCAOpen_podman | ||
|
|
||
|
|
||
| # SYSTEM-SPECIFIC PARAMETERS (please change to match your local setup): | ||
| project_root_host: $CODEBASE/OpenScPCA-analysis/ | ||
|
|
||
|
|
||
| # PATHS AS VISIBLE WITHIN RSTUDIO (should not be changed): | ||
| project_root: /home/rstudio | ||
| tmp_root: /home/rstudio/mnt_tmp/ | ||
|
|
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,23 @@ | ||
| gene_symbol,ENSEMBL_ID,cell_class,cell_type,DOI,comment | ||
| WT1,ENSG00000184937,malignant,cancer_cell,10.1242/dev.153163,Tumor_suppressor_WT1_is_lost_in_some_WT_cells | ||
| IGF2,ENSG00000167244,malignant,cancer_cell,10.1038/ng1293-408,NA | ||
| TP53,ENSG00000141510,malignant,anaplastic,10.1158/1078-0432.CCR-16-0985,Might_also_be_in_small_non_anaplastic_subset | ||
| MYCN,ENSG00000134323,malignant,anaplastic,10.18632/oncotarget.3377,Also_in_non_anaplastic_poor_outcome | ||
| MAX,ENSG00000125952,malignant,anaplastic,10.1016/j.ccell.2015.01.002,Also_in_non_anaplastic_poor_outcome | ||
| SIX1,ENSG00000126778,malignant,blastema,10.1016/j.ccell.2015.01.002,NA | ||
| SIX2,ENSG00000170577,malignant,blastema,10.1016/j.ccell.2015.01.002,NA | ||
| CITED1,ENSG00000125931,malignant,blastema,10.1593/neo.07358,Also_in_embryonic_kidney | ||
| PTPRC,ENSG00000081237,immune,NA,10.1101/gr.273300.120,NA | ||
| CD68,ENSG00000129226,immune,myeloid,10.1186/1746-1596-7-12,NA | ||
| CD163,ENSG00000177575,immune,macrophage,10.1186/1746-1596-7-12,NA | ||
| VWF,ENSG00000110799,endothelium,endothelium,10.1134/S1990747819030140,NA | ||
| CD3E,ENSG00000198851,immune,T_cell,10.1101/gr.273300.120,NA | ||
| MS4A1,ENSG00000156738,immune,B_cell,10.1101/gr.273300.120,NA | ||
| FOXP3,ENSG00000049768,immune,T_cell,10.1101/gr.273300.120,Treg | ||
| CD4,ENSG00000010610,immune,T_cell,10.1101/gr.273300.120,NA | ||
| CD8A,ENSG00000153563,immune,T_cell,10.1101/gr.273300.120,NA | ||
| EPCAM,ENSG00000119888,NA,epithelial,10.1016/j.stemcr.2014.05.013,epithelial_malignant_and_non_malignant | ||
| NCAM1,ENSG00000149294,malignant,blastema,10.1016/j.stemcr.2014.05.013,might_also_be_expressed_in_non_malignant | ||
| PODXL,ENSG00000128567,non-malignant,podocyte,10.1016/j.stem.2019.06.009,NA | ||
| COL6A3,ENSG00000163359,malignant,mesenchymal,10.2147/OTT.S256654,might_also_be_expressed_in_non_malignant_stroma | ||
| THY1,ENSG00000154096,malignant,mesenchymal,10.1093/hmg/ddq042,might_also_be_expressed_in_non_malignant_stroma |
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,6 @@ | ||
| alteration,gain_loss,cell_class,cell_type,DOI,PMID,comment | ||
| 11p13,loss,malignant,NA,10.1242/dev.153163,NA,NA | ||
| 11p15,loss,malignant,NA,10.1128/mcb.9.4.1799,NA,NA | ||
| 16q,loss,malignant,NA,NA,1317258,Associated_with_relapse | ||
| 1p,loss,malignant,NA,NA,8162576,Associated_with_relapse | ||
| 1q,gain,malignant,NA,10.1016/S0002-9440(10)63982-X,NA,Associated_with_relapse |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The workflow added with
.github/workflows/docker_cell-type-wilms-tumor-06.ymlpassed: https://github.com/AlexsLemonade/OpenScPCA-analysis/actions/runs/10238728443/job/28323567327So, now we can be confident that this builds successfully 🥳
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is a great news, thank you!!!
Should we wait a more complete Docker image before pushing it?
I would like to adapt the notebook with you comment (enrichment analysis of some gene sets) before filling the next pull request, with a docker image (renv.lock) that should be more complete.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There's no need to wait – incremental changes are okay and expected! As you add R packages you need, I expect the
renv.lockfile to change, which means the Docker image will get rebuilt with the new R packages installed and pushed to ECR. That is to say that the workflow is set up to accommodate incremental changes!There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
great !