Merge pull request #5 from chrarnold/main

Updated GRaNIE resources

dockerfiles/granie/Dockerfile

@@ -0,0 +1,39 @@
+# Use the base image
+FROM  bioconductor/bioconductor_docker:devel-R-4.4.1
+
+
+# Install required dependencies for the R packages
+#RUN apt-get update && apt-get install -y \
+#    libcurl4-openssl-dev \
+#    libxml2-dev \
+#    libssl-dev \
+#    libcairo2-dev 
+#    libxt-dev \
+#    libopenblas-dev
+
+
+#Install R packages
+RUN R -e "install.packages(c('devtools'))"
+
+# Set the working directory
+WORKDIR /workspace
+
+# Default command
+CMD ["R"]
+
+RUN R -e "remotes::install_version('Matrix', version = '1.6-3')"
+
+#Version 1.9.4 of GRaNIE
+RUN R -e "devtools::install_gitlab('grp-zaugg/GRaNIE@855bf3def33ad7af9353db79c7e21c9279035fb8', host = 'git.embl.de', subdir = 'src/GRaNIE', dependencies = TRUE, upgrade = 'never')"
+
+#Version 0.2.1 of GRaNIEverse
+RUN R -e "devtools::install_gitlab('grp-zaugg/GRaNIEverse@3224b042f25f0085eeaf9194042bcb89103ea962', host = 'git.embl.de', upgrade = 'never', dependencies = TRUE)"
+
+#There seems to be an incompatibility with the Matrix and irlba package, downgrading Matrix seems to work
+
+RUN R -e "install.packages('irlba',type='source')"
+
+RUN R -e "BiocManager::install(c('BSgenome.Hsapiens.UCSC.hg38', 'EnsDb.Hsapiens.v86', 'EnsDb.Mmusculus.v79', 'BSgenome.Mmusculus.UCSC.mm39'))"
+
+
+

src/methods/multi_omics/granie/config.novsh.yaml

@@ -0,0 +1,138 @@
+__merge__: ../../../api/comp_method.yaml
+
+
+functionality:
+  name: GRaNIE
+  namespace: "grn_methods"
+  info:
+    label: GRaNIE
+    summary: "GRN inference using GRaNIE"
+    description: |
+      GRN inference using GRaNIE
+    documentation_url: https://grp-zaugg.embl-community.io/GRaNIE/
+  arguments:
+  - name: --file_rna
+    type: file
+    required: true
+    direction: input
+    description: "Path to the RNA data file (e.g., rna.rds)."
+
+  - name: --file_atac
+    type: file
+    required: true
+    direction: input
+    description: "Path to the ATAC data file (e.g., atac.rds)."
+
+  - name: --temp_dir
+    type: directory
+    required: false
+    direction: input
+    default: "output/granie/"
+    description: "Temporary directory for output files."
+
+  - name: --preprocessing_clusteringMethod
+    type: integer
+    required: false
+    default: 1
+    direction: input
+    description: "Seurat::FindClusters: Clustering method to use (1 = original Louvain algorithm, 2 = Louvain algorithm with multilevel refinement, 3 = SLM algorithm, 4 = Leiden algorithm)."
+
+  - name: --preprocessing_clusterResolution
+    type: float
+    required: true
+    direction: input
+    description: "Resolution for clustering, typically between 5 and 20."
+
+  - name: --preprocessing_SCT_nDimensions
+    type: integer
+    required: false
+    direction: input
+    default: 50
+    description: "Number of dimensions for SCT, default is 50."
+
+  - name: --genomeAssembly
+    type: string
+    required: true
+    direction: input
+    description: "Genome assembly version (e.g., hg38). Currently, hg38 and mm10 is supported."
+
+  - name: --GRaNIE_corMethod
+    type: string
+    required: false
+    default: "spearman"
+    direction: input
+    description: "Correlation method used in GRaNIE (e.g., 'spearman')."
+
+  - name: --GRaNIE_includeSexChr
+    type: boolean
+    required: false
+    default: true
+    direction: input
+    description: "Include sex chromosomes in analysis."
+
+  - name: --GRaNIE_promoterRange
+    type: integer
+    required: false
+    default: 250000
+    direction: input
+    description: "Range in base pairs for maximum distance of peak-gene connections (default is 250000)."
+
+  - name: --GRaNIE_TF_peak.fdr.threshold
+    type: float
+    required: false
+    default: 0.2
+    direction: input
+    description: "FDR threshold for TF-peak connections (default is 0.2)."
+
+  - name: --GRaNIE_peak_gene.fdr.threshold
+    type: float
+    required: false
+    default: 0.2
+    direction: input
+    description: "FDR threshold for peak-gene connections (default is 0.2)."
+
+  - name: --GRaNIE_nCores
+    type: integer
+    required: true
+    direction: input
+    description: "Number of cores to use for computation (default is 4)."
+
+  - name: --peak_gene
+    type: file
+    required: false
+    direction: output
+    description: "Path to the peak-gene output file (e.g., peak_gene.csv). Not yet implemented."
+
+  - name: --prediction
+    type: file
+    required: true
+    direction: output
+    description: "Path to the prediction output file (e.g., prediction.csv)."
+
+  - name: --useWeightingLinks
+    type: boolean
+    required: false
+    default: false
+    direction: input
+    description: "Flag to indicate whether to use weighting links in analysis."
+
+  - name: --forceRerun
+    type: boolean
+    required: false
+    default: false
+    direction: input
+    description: "Flag to force rerun of the analysis regardless of existing results."
+
+  resources:
+    - type: r_script
+      path: script.R
+
+
+platforms:
+  - type: docker
+    image: chrarnold84/granieverse:latest
+
+  - type: native
+  - type: nextflow
+    directives:
+      label: [midtime,midmem,midcpu]

src/methods/multi_omics/granie/script.R

@@ -14,25 +14,25 @@ library(BSgenome.Mmusculus.UCSC.mm39)
 library(dplyr)
 
 ## VIASH START
-par <- list(
-  file_rna = "resources_test/grn-benchmark/multiomics_r/rna.rds",
-  files_atac = "resources_test/grn-benchmark/multiomics_r/atac.rds",
-  temp_dir =  "output/granie/",
-  preprocessing_clusteringMethod = 1, # Seurat::FindClusters: (1 = original Louvain algorithm, 2 = Louvain algorithm with multilevel refinement, 3 = SLM algorithm, 4 = Leiden algorithm)
-  preprocessing_clusterResolution = 14, # Typically between 5 and 20
-  preprocessing_SCT_nDimensions = 50, # Default 50
-  genomeAssembly = "hg38",
-  GRaNIE_corMethod = "spearman",
-  GRaNIE_includeSexChr = TRUE,
-  GRaNIE_promoterRange = 250000,
-  GRaNIE_TF_peak.fdr.threshold = 0.2,
-  GRaNIE_peak_gene.fdr.threshold = 0.2,
-  GRaNIE_nCores = 4,
-  peak_gene = "output/granie/peak_gene.csv", # not yet implemented, should I?
-  prediction= "output/granie/prediction.csv",
-  useWeightingLinks = FALSE,
-  forceRerun = FALSE
-)
+# par <- list(
+#   file_rna = "resources_test/grn-benchmark/multiomics_r/rna.rds",
+#   file_atac = "resources_test/grn-benchmark/multiomics_r/atac.rds",
+#   temp_dir =  "output/granie/",
+#   preprocessing_clusteringMethod = 1, # Seurat::FindClusters: (1 = original Louvain algorithm, 2 = Louvain algorithm with multilevel refinement, 3 = SLM algorithm, 4 = Leiden algorithm)
+#   preprocessing_clusterResolution = 14, # Typically between 5 and 20
+#   preprocessing_SCT_nDimensions = 50, # Default 50
+#   genomeAssembly = "hg38",
+#   GRaNIE_corMethod = "spearman",
+#   GRaNIE_includeSexChr = TRUE,
+#   GRaNIE_promoterRange = 250000,
+#   GRaNIE_TF_peak.fdr.threshold = 0.2,
+#   GRaNIE_peak_gene.fdr.threshold = 0.2,
+#   GRaNIE_nCores = 4,
+#   peak_gene = "output/granie/peak_gene.csv", # not yet implemented, should I?
+#   prediction= "output/granie/prediction.csv",
+#   useWeightingLinks = FALSE,
+#   forceRerun = FALSE
+# )
 
 print(par)
 # meta <- list(