def diffexp(branch: "CellBranch", run_name: str):
# TODO: refactor both diffexp versions into `_get_diffexp_args`
input_metadata_path = branch.get_temp_meta_path(run_name)
run_spec = branch.spec[run_name]
params = run_spec.params
process_run = branch[run_name]
output_diffexp_path = process_run.path_map["diffexp_result"]
root_dir = str(branch.root)
spec_str = branch.spec.shell_str
groups = branch[run_name].branch.meta["partition_code"].unique().astype(
"O")
if len(groups) != 2:
raise ValueError(
f"Exactly two groups required for diffexp. Got: {groups}")
ident1 = groups.min()
ident2 = groups.max()
groupby = "partition_code"
arg_list = [
input_metadata_path,
output_diffexp_path,
root_dir,
spec_str,
params["test"],
params["logfc_thresh"],
ident1,
ident2,
groupby,
N_CPUS,
# R_FUNCTIONS_FILEPATH,
]
run_r_script_logged(branch, R_DIFFEXP_SCRIPT, arg_list, run_name)
def reduce(branch: "CellBranch", run_name: str):
input_metadata_path = branch.get_temp_meta_path(run_name)
run_spec = branch.spec[run_name]
params = run_spec.params
process_run = branch[run_name]
input_rds_path = process_run.path_map_prior["rna_r"]
output_embeddings_path = process_run.path_map["pca_embeddings"]
output_loadings_path = process_run.path_map["pca_loadings"]
output_stdev_path = process_run.path_map["pca_stdev"]
npcs = params["pca_npcs"]
# r_functions_filepath = R_FUNCTIONS_FILEPATH
arg_list = [
input_metadata_path,
input_rds_path,
output_embeddings_path,
output_loadings_path,
output_stdev_path,
npcs,
# r_functions_filepath,
]
run_r_script_logged(branch, R_PCA_SCRIPT, arg_list, run_name)
meta = run_umap(
output_embeddings_path,
n_neighbors=params["umap_n_neighbors"],
min_dist=params["umap_min_dist"],
n_components=params["umap_n_components"],
metric=params["umap_metric"],
)
meta.index = branch.meta.index
output_meta_path = process_run.path_map["meta"]
meta.to_csv(output_meta_path, sep="\t")
def markers(branch: "CellBranch", run_name: str):
input_metadata_path = branch.get_temp_meta_path(run_name)
run_spec = branch.spec[run_name]
params = run_spec.params
process_run = branch[run_name]
output_markers_path = process_run.path
root_dir = str(branch.root)
spec_str = branch.spec.shell_str
cluster_counts = branch.meta["cluster_id"].value_counts()
deficient_clusters = cluster_counts[cluster_counts < 2].index.tolist()
if deficient_clusters:
raise ValueError(
f"Deficient clusters {deficient_clusters} with fewer than 3 "
f"cells. Adjust clustering parameters.\n{cluster_counts}")
arg_list = [
input_metadata_path,
output_markers_path,
root_dir,
spec_str,
params["logfc_thresh"],
params["test"],
N_CPUS,
# R_FUNCTIONS_FILEPATH,
]
run_r_script_logged(branch, R_MARKERS_SCRIPT, arg_list, run_name)
def normalize(branch: "CellBranch", run_name: str):
"""
Performs:
- cell filtering by `min_genes`, `max_genes`, and `perc_mito_cutoff`
- gene filtering by `min_cells` expressing
- normalization via either:
- seurat default
- sctransform
Params:
min_genes (int):
max_genes (int):
min_cells (int):
perc_mito_cutoff (int, float):
method (str): from {"seurat_default", "sctransform"}
nfeatures (int): (seurat_default only)
"""
input_metadata_path = branch.get_temp_meta_path(run_name)
# TODO: add a root filepaths lookup
run_spec = branch.spec[run_name]
params = run_spec.params
process_run = branch[run_name]
input_rds_path = process_run.path_map_prior["rna_r"]
output_rds_path = process_run.path_map["rna_r"]
min_genes = params["min_genes"]
max_genes = params["max_genes"]
min_cells = params["min_cells"]
perc_mito_cutoff = params["perc_mito_cutoff"]
method = params["method"]
arg_list = [
input_metadata_path,
input_rds_path,
output_rds_path,
min_genes,
max_genes,
min_cells,
perc_mito_cutoff,
# R_FUNCTIONS_FILEPATH,
]
if method == "sctransform":
output_corrected_umi_path = process_run.path_map["corrected_umi"]
output_pearson_residual_path = process_run.path_map["pearson_residual"]
arg_list += [output_corrected_umi_path, output_pearson_residual_path]
r_normalize_script = R_SCTRANSFORM_SCRIPT
elif method == "seurat_default":
verbose = True
verbose = str(verbose).upper()
nfeatures = params["nfeatures"]
arg_list += [verbose, nfeatures]
r_normalize_script = R_SEURAT_DEFAULT_NORM_SCRIPT
else:
raise ValueError(
f"Invalid normalization method: {method}. Use 'sctransform' or 'seurat_default'"
)
run_r_script_logged(branch, r_normalize_script, arg_list, run_name)
def cluster(branch: "CellBranch", run_name: str):
input_metadata_path = branch.get_temp_meta_path(run_name)
# TODO: add a root filepaths lookup
run_spec = branch.spec[run_name]
params = run_spec.params
process_run = branch[run_name]
output_clusters_path = process_run.path_map["meta"]
# TODO: may actually need way to pass spec through
root_dir = str(branch.root)
spec_str = branch.spec.shell_str
arg_list = [
input_metadata_path,
output_clusters_path,
root_dir,
spec_str,
params["num_pcs"],
params["res"],
params["eps"],
# R_FUNCTIONS_FILEPATH,
]
run_r_script_logged(branch, R_CLUSTER_SCRIPT, arg_list, run_name)
def plot_cell_cycle_scoring_scat(branch: "CellBranch", r_script: str,
args: list, **kwargs):
run_r_script_logged(branch, r_script, args, "plot_cell_cycle_scoring_scat")
def plot_umap_embeddings_scat_r(branch: "CellBranch", r_script: str,
args: list, **kwargs):
run_r_script_logged(branch, r_script, args, "plot_umap_embeddings_scat")
def plot_pca_loadings_scat(branch: "CellBranch", r_script: str, args: list,
**kwargs):
run_r_script_logged(branch, r_script, args, "plot_pca_loadings_scat")
def plot_pca_elbow_curv(branch: "CellBranch", r_script: str, args: list,
**kwargs):
run_r_script_logged(branch, r_script, args, "plot_pca_elbow_curv")
def plot_perc_hsp_per_cell_vln_r(branch: "CellBranch", r_script: str,
args: list, **kwargs):
run_r_script_logged(branch, r_script, args, "plot_perc_hsp_per_cell_vln_r")
def plot_umis_vs_perc_ribo_scat_r(branch: "CellBranch", r_script: str,
args: list, **kwargs):
run_r_script_logged(branch, r_script, args, "plot_umis_vs_perc_ribo_scat")
def plot_marker_genes_per_cluster_bar(branch: "CellBranch", r_script: str,
args: list, **kwargs):
args.append(branch.current_path)
run_r_script_logged(branch, r_script, args,
"plot_marker_genes_per_cluster_bar")
def plot_umis_per_barcode_rank_curv(branch: "CellBranch", r_script: str,
args: list, **kwargs):
run_r_script_logged(branch, r_script, args, branch.current_process)
def plot_highest_exprs_dens(branch: "CellBranch", r_script: str, args: list,
**kwargs):
run_r_script_logged(branch, r_script, args, "plot_highest_exprs_dens")
