Parallel process clusters and updates to cytetype v0.13.0 (#35)

* prod link update

* Update .gitignore

* n_parallel_clus doc

* n_parallel_clus_validate

* add ontology term name

* update llm providers

* fix parallel clus class

* save job_details

* Update cytetype.R

* n_parallel_clus inputdata

* Update schema.R

* Update .gitignore

* Update schema.R

* Update cytetype.R

* Update DESCRIPTION

* Increment version number to 0.2.0

Author: suu-yi

.gitignore

@@ -9,3 +9,5 @@
 revdep/
 inst/doc
 CyteTypeR.Rproj
+query.json
+*.json

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: CyteTypeR
 Title: CyteType for R
-Version: 0.1.5
+Version: 0.2.0
 Description: CyteTypeR is the R version of CyteType python package.'
 Authors@R: 
     person("Nygen Analytics AB", , ,"contact@nygen.io", role = c("aut", "cre"))

R/cytetype.R

@@ -154,6 +154,8 @@ PrepareCyteTypeR <- function(obj,
 #'   (in seconds). Default is 30.
 #' @param timeout_seconds Integer specifying maximum wait time for job completion
 #'   (in seconds). Default is 1200 (20 minutes).
+#' @param n_parallel_clusters Number of parallel requests to make to the model. Maximum is 50. Note than high values can lead to rate limit errors.
+
 #' @param api_url Optional character string specifying custom API URL.
 #'   If `NULL`, uses default URL. Default is `NULL`.
 #' @param save_query Logical indicating whether to save the query as JSON file.
@@ -213,6 +215,7 @@ CyteTypeR <- function(obj,
                       results_prefix = "cytetype",
                       poll_interval_seconds = 30,
                       timeout_seconds = 1200,
+                      n_parallel_clusters = 2,
                       api_url = NULL,
                       save_query = TRUE,
                       query_filename = "query.json",
@@ -222,6 +225,7 @@ CyteTypeR <- function(obj,
   api_url <- api_url %||% .get_default_api_url()
   prepped_data$studyInfo <- study_context %||% ""
   prepped_data$infoTags <- metadata %||% list()
+  prepped_data$nParallelClusters <- n_parallel_clusters
 
   group_key <- prepped_data$group_key
   prepped_data$group_key <- NULL
@@ -245,21 +249,18 @@ CyteTypeR <- function(obj,
     llm_configs = llm_configs
   )
 
-  if (save_query){
-    query_for_json <- .prepare_query_for_json(query_list)
-    write_json(query_for_json, path = query_filename, auto_unbox = TRUE, pretty = TRUE)
-  }
+  # Prep query for submission
+  query_for_json <- .prepare_query_for_json(query_list)
+
   ## NA value check on all data before submitting job
 
 
   # Job submission
-
   job_id <- .submit_job(query_for_json, api_url, auth_token)
   if (is.na(job_id)) {
     stop("Job submission failed.")
   }
 
-
   # Save job details
   report_url <- file.path(api_url, 'report',job_id)
 
@@ -270,6 +271,16 @@ CyteTypeR <- function(obj,
     auth_token = auth_token
   )
 
+  if (exists("prepped_data", envir = .GlobalEnv)){
+    job_detail_fieldname <- paste0(results_prefix,'_jobDetails')
+    prepped_data[[job_detail_fieldname]] <<- job_details
+  }
+
+  # Save the query as json file if true
+  if (save_query){
+    write_json(query_for_json, path = query_filename, auto_unbox = TRUE, pretty = TRUE)
+  }
+
   # poll for results
   result <- .poll_for_results(
     job_id,
@@ -287,15 +298,19 @@ CyteTypeR <- function(obj,
     obj@misc$cytetype_results <- transformed_results
 
     ann_colname <- paste(results_prefix, group_key, sep = "_" )
-    onto_colname <- paste("cytetype_cell_ontology", group_key, sep = "_")
+    onto_colname <- paste(results_prefix, "ontologyTerm", group_key, sep = "_")
+    ontoID_colname <- paste(results_prefix, "ontologyID", group_key, sep = "_")
 
     cluster_ann_map <- setNames(transformed_results$annotation,
                                 transformed_results$clusterId)
     cluster_onto_map <- setNames(transformed_results$ontologyTerm,
                                 transformed_results$clusterId)
+    cluster_onto_id_map <- setNames(transformed_results$ontologyTermID,
+                                    transformed_results$clusterId)
 
     obj@meta.data[[ann_colname]] <- factor(cluster_ann_map[as.vector(obj@meta.data[[group_key]])])
     obj@meta.data[[onto_colname]] <- factor(cluster_onto_map[as.vector(obj@meta.data[[group_key]])])
+    obj@meta.data[[ontoID_colname]] <- factor(cluster_onto_id_map[as.vector(obj@meta.data[[group_key]])])
 
 
     return(obj)
@@ -336,6 +351,11 @@ CyteTypeR <- function(obj,
 #' @importFrom jsonlite write_json
 #' @export
 GetResults <- function(job_id = NULL){
+
+  if (is.null(job_id)){
+    stop("Please provide the job_id for your CyteType run.")
+  }
+
   tryCatch({
     # if (is.null(job_id)){
     #   job_details <- fromJSON(paste0('job_details_', job_id, '.json'))

R/schema.R

@@ -1,5 +1,16 @@
 # Valid LLM Provider Names
-LLM_PROVIDERS <- c("google", "openai", "anthropic", "groq", "mistral", "openrouter", "bedrock")
+LLM_PROVIDERS <- c("anthropic",
+                   "bedrock",
+                   "fireworks",
+                   "google",
+                   "groq",
+                   "huggingface",
+                   "mistral",
+                   "openai",
+                   "openrouter",
+                   "vertex",
+                   "xai"
+)
 
 # Valid Agent Type Names
 AGENT_TYPES <- c("contextualizer", "annotator", "reviewer", "summarizer", "clinician", "chat")
@@ -91,6 +102,13 @@ LLMModelConfig <- function(provider,
       obj_list[!sapply(obj_list, is.null)]
     })
   }
+
+  # Clear saved job_details from query for job submission
+  mask_job_details <- !endsWith(names(query_list$input_data), 'jobDetails')
+
+  query_list$input_data <- query_list$input_data[mask_job_details]
+
+
   return(query_list)
 }
 
@@ -127,7 +145,8 @@ InputData <- function(studyInfo = "",
                       clusterMetadata = list(),
                       markerGenes = list(),
                       visualizationData = NULL,
-                      expressionData = list()) {
+                      expressionData = list(),
+                      nParallelClusters = numeric()) {
 
   # Create the object
   obj <- list(
@@ -137,7 +156,8 @@ InputData <- function(studyInfo = "",
     clusterMetadata = clusterMetadata,
     markerGenes = markerGenes,
     visualizationData = visualizationData,
-    expressionData = expressionData
+    expressionData = expressionData,
+    nParallelClusters = nParallelClusters
   )
 
   # Set class
@@ -197,6 +217,13 @@ InputData <- function(studyInfo = "",
     stop("expressionData must be a named list")
   }
 
+  if (!is.numeric(obj$nParallelClusters) ||
+      obj$nParallelClusters != round(obj$nParallelClusters) ||
+      obj$nParallelClusters < 1 ||
+      obj$nParallelClusters > 50) {
+    stop("n_parallel_clusters must be an integer in range of 1 to 50")
+  }
+
   # Validate nested structure of clusterMetadata
   if (length(obj$clusterMetadata) > 0) {
     for (cluster_id in names(obj$clusterMetadata)) {
@@ -316,6 +343,9 @@ get_example_input_data <- function() {
         "Cluster2" = 2.8,
         "Cluster3" = 5.2
       )
-    )
+    ),
+
+    nParallelClusters = 5
+
   )
 }

R/seurat_helpers.R

@@ -224,7 +224,8 @@
     data.frame(
       clusterId = as.character(cluster_map[ann$clusterId] %||% ann$clusterId),
       annotation = as.character(ann$annotation %||% 'Unknown'),
-      ontologyTerm = as.character(ann$cellOntologyTerm %||% 'Unknown'),
+      ontologyTerm = as.character(ann$cellOntologyTermName %||% 'Unknown'),
+      ontologyTermID = as.character(ann$cellOntologyTerm %||% 'Unknown'),
       granularAnnotation = as.character(ann$granularAnnotation %||% ''),
       cellState = as.character(ann$cellState %||% ''),
       justification = as.character(ann$justification %||% ''),

R/zzz.R

@@ -2,7 +2,7 @@
 .onLoad <- function(libname, pkgname) {
   # Set default configuration options
   options(
-    cytetype.default.api.url = "https://nygen-labs-prod--cytetype-api.modal.run",
+    cytetype.default.api.url = "https://prod.cytetype.nygen.io",
     cytetype.default.poll.interval = 10L,
     cytetype.default.timeout = 7200L
   )
@@ -14,7 +14,7 @@
 
 # Get Default API URL
 .get_default_api_url <- function() {
-  getOption("cytetype.default.api.url", "https://nygen-labs-prod--cytetype-api.modal.run")
+  getOption("cytetype.default.api.url", "https://prod.cytetype.nygen.io")
 }
 
 # Get Default Poll Interval