Feature updates

[suu-yi]

Preprocessing

• PrepareCyteTypeR: Uses .sample_visualization_data() for visualization data

Artifacts & uploads

• New R/artifacts.R: Build normalized expression (cells × genes) from the Seurat object to vars.h5 (Optional feature metadata df)
  • cell metadata to obs.duckdb with optional config
• Upload helpers in R/client.R: .MAX_UPLOAD_BYTES (obs_duckdb 100MB, vars_h5 10GB), .upload_file_binary, .upload_obs_duckdb, .upload_vars_h5.
  • Chunked file upload and improved error logging.
• CyteTypeR():
  • upload_artifacts (default FALSE)
  • vars_h5_path, obs_duckdb_path, upload_timeout_seconds, cleanup_artifacts

API & schema

• Payload / options: Default API URL (prod), n_parallel_clusters (1–50), clientInfo (clientType, clientVersion) in input data; poll interval and timeout from options in zzz.R.
• R/schema.R: InputData and validation extended with nParallelClusters and clientInfo;
• R/zzz.R: Options cytetype.default.api.url, cytetype.default.poll.interval, cytetype.default.timeout; helpers .get_default_api_url, .get_default_poll_interval, .get_default_timeout.

Results & Seurat integration

• Results handling: override_existing_results, results_prefix; job details and annotations stored in obj@misc by prefix; GetResults(obj, job_id, results_prefix, auth_token) to read from object or fetch from API.
• R/seurat_helpers.R: .normalize_result_annotations, .store_job_details_seurat, .store_annotations_seurat, .sample_visualization_data for visualization data in prepared payload; .transform_results_seurat; annotation columns use character indexing for numeric group_key

[parashardhapola]

zzz.R can have a better name

[suu-yi]

zzz.R can have a better name

R loads files alphabetically, this ensures it loads last!