scRNA-seq_analysis

tools/AGA/AGA_from_Seurat.py

@@ -0,0 +1,93 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Wed Aug  1 15:18:12 2018
+
+@author: doru
+"""
+
+import sys
+args = sys.argv
+CWD = args[1]
+#split_cat     = args[2]
+from os import chdir
+chdir(CWD)
+
+import matplotlib; matplotlib.use('Agg');
+import scanpy.api as sc;
+import pandas as pd
+from scipy.sparse import csr_matrix
+import numpy as np
+
+sc.settings.verbosity = 3
+
+scObj = sc.read("./material/raw_data.mtx", cache = False).T
+
+# load gene names
+scObj.var_names = pd.read_csv("./material/genenames.csv").iloc[:, 1]
+
+# load cell names
+scObj.obs_names = pd.read_csv("./material/cellnames.csv").iloc[:, 1]
+
+# filter out genes present in less than 3 cells
+sc.pp.filter_genes(scObj, min_cells=3)
+
+# log-normalize the data
+scObj.raw = sc.pp.log1p(scObj, copy=True)
+sc.pp.normalize_per_cell(scObj, counts_per_cell_after=1e4)
+
+# variable genes
+filter_result = sc.pp.filter_genes_dispersion(
+        scObj.X, min_mean=0.0125, max_mean=3, min_disp=0.5)
+# subset data on variable genes
+scObj = scObj[:, filter_result.gene_subset]
+# not sure?
+sc.pp.log1p(scObj)
+
+# scale the data
+sc.pp.scale(scObj, max_value=10)
+
+# run pca
+sc.tl.pca(scObj)
+
+# compunte neighborhood graph
+sc.pp.neighbors(scObj, n_neighbors = 15, n_pcs = 20, knn = True, random_state = 10, method = "gauss")
+
+# add cell labels
+cell_labels = pd.read_csv("./material/cell_labels.csv", index_col = 0)
+
+scObj.obs["cell_labels"] = cell_labels
+
+# run aga
+sc.tl.paga(scObj, groups = "cell_labels")
+
+# save the scipy paga graph to disk for comparison to the plot generated by ggplot - for trouble shooting
+
+sc.pl.paga(scObj,  save = "_ugly_scanpy_plot.pdf", show = True, edge_width_scale = .4, solid_edges = "connectivities", layout="fa")
+
+#sc.pl.paga(scObj,  save = "ugly_scanpy_plot.pdf", show = False, edge_width_scale = .4, solid_edges = "connectivities",
+#           layout = "fa") 
+
+#sc.pl.paga(scObj,  save = "_ugly_scanpy_plot.pdf", show = True, edge_width_scale = .4, solid_edges = "connectivities", layout = "fa")
+
+#sc.pl.paga(scObj,  save = "{split_cat}_ugly_scanpy_plot.pdf".format(split_cat=split_cat), show = True, edge_width_scale = .4, solid_edges = "connectivities", layout = "fa")
+
+#sc.pl.paga(scObj,  save = "{split_cat}_ugly_scanpy_plot.pdf".format(split_cat=split_cat),
+#           show = True, edge_width_scale = .4, solid_edges = "connectivities",
+#           layout = "fa") # layout = "fa"
+
+# prepare the output and save it to disk
+cell_cats = list(scObj.obs["cell_labels"].cat.categories)
+population_size = cell_labels["Labels"].value_counts()
+population_size = population_size[cell_cats].values
+
+connectivities = np.array(csr_matrix.todense(scObj.uns["paga"]["connectivities"]), dtype = "float64")
+connectivities[connectivities < .05] = 0.0
+
+#connectivities[connectivities < .1] = 0.0
+connectivities = pd.DataFrame(connectivities, columns = cell_cats, index = cell_cats)
+connectivities.to_csv("./AGA_folder/connectivities.csv", index = True, header = True)
+
+coordinates = pd.DataFrame(scObj.uns["paga"]["pos"], columns = ["X", "Y"], index = cell_cats)
+coordinates["Size"] = pd.Series(population_size, coordinates.index)
+coordinates.to_csv("AGA_folder/coordinates.csv", index = True, header = True)