diff --git a/src/methods/lmds_irlba_rf/config.vsh.yaml b/src/methods/lmds_irlba_rf/config.vsh.yaml
deleted file mode 100644
index 8f85382..0000000
--- a/src/methods/lmds_irlba_rf/config.vsh.yaml
+++ /dev/null
@@ -1,39 +0,0 @@
-__merge__: ../../api/comp_method.yaml
-name: lmds_irlba_rf
-label: LMDS + IRLBA + RF
-summary: A random forest regression using LMDS of modality 1 to predict a PCA embedding of modality 2, which is then reversed to predict the original modality 2.
-description: |
-  A random forest regression using LMDS of modality 1 to predict a PCA embedding of modality 2, which is then reversed to predict the original modality 2.
-references:
-  doi: 10.1101/2022.04.11.487796
-info:
-  documentation_url: https://github.com/openproblems-bio/openproblems-v2/tree/main/src/tasks/predict_modality/methods #/lmds_irlba_rf
-  repository_url: https://github.com/openproblems-bio/openproblems-v2
-  preferred_normalization: log_cp10k
-arguments:
-  - name: "--distance_method"
-    type: "string"
-    default: "pearson"
-    description: The distance method to use. Possible values are euclidean, pearson, spearman and others.
-  - name: "--n_pcs"
-    type: "integer"
-    default: 20
-    description: Number of principal components to use.
-  - name: "--n_trees"
-    type: "integer"
-    default: 500
-    description: Number of trees to use.
-resources:
-  - type: r_script
-    path: script.R
-engines:
-  - type: docker
-    image: openproblems/base_r:1.0.0
-    setup:
-      - type: r
-        cran: [lmds, ranger, pbapply, irlba]
-runners:
-  - type: executable
-  - type: nextflow
-    directives:
-      label: [hightime, highmem, highcpu]
\ No newline at end of file
diff --git a/src/methods/lmds_irlba_rf/script.R b/src/methods/lmds_irlba_rf/script.R
deleted file mode 100644
index 00e1ac7..0000000
--- a/src/methods/lmds_irlba_rf/script.R
+++ /dev/null
@@ -1,93 +0,0 @@
-cat("Loading dependencies\n")
-requireNamespace("anndata", quietly = TRUE)
-requireNamespace("pbapply", quietly = TRUE)
-library(Matrix, warn.conflicts = FALSE, quietly = TRUE)
-
-## VIASH START
-path <- "resources_test/task_predict_modality/openproblems_neurips2021/bmmc_multiome/normal/"
-par <- list(
-  input_train_mod1 = paste0(path, "train_mod1.h5ad"),
-  input_test_mod1 = paste0(path, "test_mod1.h5ad"),
-  input_train_mod2 = paste0(path, "train_mod2.h5ad"),
-  output = "output.h5ad",
-  n_pcs = 20L,
-  n_trees = 50L
-)
-meta <- list(name = "foo")
-## VIASH END
-
-n_cores <- parallel::detectCores(all.tests = FALSE, logical = TRUE)
-
-cat("Reading mod1 files\n")
-input_train_mod1 <- anndata::read_h5ad(par$input_train_mod1)
-input_test_mod1 <- anndata::read_h5ad(par$input_test_mod1)
-
-dataset_id <- input_train_mod1$uns[["dataset_id"]]
-
-cat("Performing DR on the mod1 values\n")
-dr <- lmds::lmds(
-  rbind(input_train_mod1$layers[["normalized"]], input_test_mod1$layers[["normalized"]]), 
-  ndim = par$n_pcs,
-  distance_method = par$distance_method
-)
-# alternative:
-# pr_out <- irlba::prcomp_irlba(
-#   rbind(input_train_mod1$layers[["normalized"]], input_test_mod1$layers[["normalized"]]),
-#   n = par$n_pcs
-# )
-# dr <- pr_out$x
-
-# split up dr data
-ix <- seq_len(nrow(input_train_mod1))
-dr_train <- as.data.frame(dr[ix, , drop = FALSE])
-dr_test <- as.data.frame(dr[-ix, , drop = FALSE])
-dr_train <- dr[ix, , drop = FALSE]
-dr_test <- dr[-ix, , drop = FALSE]
-
-rm(input_train_mod1, input_test_mod1)
-gc()
-
-
-cat("Reading mod2 files\n")
-X_mod2 <- anndata::read_h5ad(par$input_train_mod2)$layers[["normalized"]]
-prcomp_mod2 <- irlba::prcomp_irlba(X_mod2, n = par$n_pcs)
-dr_mod2 <- prcomp_mod2$x
-
-cat("Predicting for each column in modality 2\n")
-pred_drs <- pbapply::pblapply(
-  seq_len(ncol(dr_mod2)),
-  function(i) {
-    y <- dr_mod2[, i]
-    uy <- unique(y)
-    if (length(uy) > 1) {
-      rf <- ranger::ranger(
-        x = dr_train,
-        y = y,
-        num.trees = par$n_trees,
-        num.threads = n_cores
-      )
-      stats::predict(rf, dr_test)$prediction
-    } else {
-      rep(uy, nrow(dr_test))
-    }
-  }
-)
-
-cat("Creating outputs object\n")
-pred_dr <- Matrix::Matrix(do.call(cbind, pred_drs), sparse = TRUE)
-prediction <- pred_dr %*% t(prcomp_mod2$rotation)
-rownames(prediction) <- rownames(dr_test)
-colnames(prediction) <- colnames(X_mod2)
-
-out <- anndata::AnnData(
-  layers = list(normalized = as(prediction, "CsparseMatrix")),
-  shape = dim(prediction),
-  uns = list(
-    dataset_id = dataset_id,
-    method_id = meta$name
-  )
-)
-
-
-cat("Writing predictions to file\n")
-zzz <- out$write_h5ad(par$output, compression = "gzip")
diff --git a/src/workflows/run_benchmark/config.vsh.yaml b/src/workflows/run_benchmark/config.vsh.yaml
index 65f222e..54fc037 100644
--- a/src/workflows/run_benchmark/config.vsh.yaml
+++ b/src/workflows/run_benchmark/config.vsh.yaml
@@ -69,7 +69,6 @@ dependencies:
   - name: methods/knnr_py
   - name: methods/knnr_r
   - name: methods/lm
-  - name: methods/lmds_irlba_rf
   - name: methods/guanlab_dengkw_pm
   - name: metrics/correlation
   - name: metrics/mse
diff --git a/src/workflows/run_benchmark/main.nf b/src/workflows/run_benchmark/main.nf
index a6dba16..133d6e5 100644
--- a/src/workflows/run_benchmark/main.nf
+++ b/src/workflows/run_benchmark/main.nf
@@ -14,7 +14,6 @@ methods = [
   knnr_py,
   knnr_r,
   lm,
-  lmds_irlba_rf,
   guanlab_dengkw_pm
 ]