Generators

Generator.omics

Flow2Spatial.generator.omics(adata, mask, dir_run='./save_environ')

Transfer molecular distribution from spatial omics as training data.

The first input is the spatial distribution of moleculars from spatial omics (in the format of anndata). And the second inputs a bool matrix, showing whether tissue slice is palced in certain pixel. In default, output file will locate in the directory “./save_environ”.

Corresponding adata and mask locate at https://github.com/bioinfo-biols/Flow2Spatial/tree/main/tests under the names Gut_reference.h5ad and mask.

Generator.histology

Flow2Spatial.generator.histology(line_row, line_col, mask, segments, channel_intensity, radius=[0.5, 0.5], Hcoordinate=None, dir_run='./save_environ')

Transfer histological information as training data.

The first two inputs (line_row, line_col) are the parameters of the line equation in the parallel-flow projection of the first and last slice, which is placed by microfluidics chip. The input format should be dict, with the key as ‘line1’ ‘line2’ … ‘lineN’ and the value as the (a, b, c) of the line equation ax + by + c = 0, hence {‘line1’: (a1, b1, c1), … ‘lineN’: (aN, bN, cN)}. The third inputs a bool matrix, showing whether tissue slice is palced in certain pixel. After that, a list of histological clusters segments and MS intensity in each channel are needed for the function.

Corresponding mask and channel_intensity locate at https://github.com/bioinfo-biols/Flow2Spatial/tree/main/tests under the names mask and df_pro_gut.csv. And segments can be generated from adata with following code:

from anndata import read_h5ad
adata = read_h5ad('./Gut_reference.h5ad')
import pickle
with open('./mask', 'rb') as handle:
    mask = pickle.load(handle)
segments = F2S.transfer_masks(adata, mask, list_s=['Cluster1', 'Cluster2', 'Cluster3', 'Cluster4', 'Cluster5', 'Cluster6', 'Cluster7', 'Cluster8', 'Cluster9'])

Moreover, radius and Hcoordinate means the radius and orignal coordinate of each point in the parallel-flow projection. If Hcoordinate is not None, it requires the coordinate based on the histology image. For the gut reconstruction, corresponding Hcoordinate locates at https://github.com/bioinfo-biols/Flow2Spatial/tree/main/tests .

In default, output file will locate in the directory “./save_environ”.

Generator.random

Flow2Spatial.generator.random(input_type=['omics', 'histology'], times=20000, dir_run='./save_environ')

Imporve the diversity of spatial distribution in the training set.

The first parameter represents which reference data will be used as the source for the random generator. We can specify [‘histology’] as the only source or both [‘omics’, ‘histology’], which is the default. We can set parameter times to control the number of samples generated. Default is 20,000. In default, output file will locate in the directory “./save_environ”.