Quick Demo: Reference-free spatial omics simulation

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

import simspace as ss

# Step 1: Generate random parameters for the simulation. The random parameters will define the characteristics of the spatial omics simulation.
# Here, we generate parameters for 3 groups (niches) and 8 states (cell types), with a fixed seed for reproducibility.
# The parameters include information about the spatial structure, such as the number of groups, states, and other simulation settings.

params = ss.util.generate_random_parameters(
    n_group=3,
    n_state=8,
    seed=42,
)

# Step 2: Create a spatial simulation using the generated parameters.
# The shape of the simulation is set to (100, 100), meaning a 100x100 grid.
# The custom_neighbor argument specifies the spatial connectivity, using Manhattan distance offsets for neighbors.
# One can check the spatial connectivity by sim.neighborhood
# The seed is set to 42 for reproducibility.
sim = ss.util.sim_from_params(
    parameters=params,
    shape=(100, 100),
    custom_neighbor=ss.spatial.generate_offsets(3, method='manhattan'),
    seed=42,
)

# Step 3: Visualize the spatial simulation.
# The simulation can be visualized using the plot_niche method, which shows the spatial distribution of different niches.
sim.plot_niche(figsize=(5, 5), dpi=100)
# The plot method can be used to visualize the entire simulation, including the spatial distribution of states.
sim.plot(figsize=(5, 5), dpi=100, size=14)

# Step 4 (Optional): Simulate omics data based on the spatial simulation.
# The create_omics method generates omics data with specified parameters.
# Here, we create 100 genes with a background ratio of 0.6 and a ligand-receptor ratio of 0.2.
# The background and marker parameters define the distribution of gene expression levels.
# The spatial argument is set to False, indicating that the omics data is not spatially dependent.
# The generated omics data can be used for further analysis or visualization.
sim.create_omics(
    n_genes=100,
    bg_ratio=0.6,
    lr_ratio=0.2,
    bg_param = (1, 0.5),
    marker_param = (5, 1.6),
    spatial=False)

sim.gene_meta.head()

	GeneID	Marker	LRindex	Type_0	Type_1	Type_2	Type_3	Type_4	Type_5	Type_6	Type_7
0	0	6	-1	0.026426	0.163313	1.194430	0.136930	0.078265	0.336136	12.154214	0.138572
1	1	-1	-1	0.651374	0.229261	0.500252	0.501919	0.383693	0.047315	0.901822	0.193405
2	2	-1	-1	0.103216	0.020815	0.446889	0.565926	0.008363	0.358815	0.128412	0.518062
3	3	-1	-1	0.095802	0.587106	0.244479	1.380172	0.073972	0.208566	0.060222	1.293095
4	4	2	39	1.049167	0.149164	10.378855	0.849742	0.405066	0.377140	0.138439	0.048863

ss.plot.plot_gene(
    sim.meta, sim.omics['Gene_0'], size=14,
    figsize=(5,5), dpi=100, title='Gene 0 Expression',
    )