spatialtis_core#

Subpackages#

Submodules#

Package Contents#

class spatialtis_core.CellCombs(types: List[str], order: bool = False)#

Profile cell-cell interaction using permutation test

Parameters

  • types – All the type of cells in your research

  • order – bool (False); If False, A->B and A<-B is the same

bootstrap(types: List[str], neighbors: spatialtis_core.types.Neighbors, labels: spatialtis_core.types.Labels, times: int = 1000, pval: float = 0.05, method: str = 'pval') → List[Tuple[str, str, float]]#

Bootstrap functions

1.0 means association, -1.0 means avoidance, 0.0 means insignificance.

Parameters

  • types – The type of all the cells

  • neighbors – List of neighbors

  • labels – List of labels

  • times – How many times to perform bootstrap

  • pval – The threshold of p-value

  • method – ‘pval’ or ‘zscore’

Returns

List of tuples, eg.(‘a’, ‘b’, 1.0), the type a and type b has a relationship as association

spatialtis_core.bbox_neighbors(bbox: List[spatialtis_core.types.BoundingBox], labels: spatialtis_core.types.Labels, expand: float = - 1.0, scale: float = 1.3) → List[List[int]]#

Get neighbors for each bounding box

Parameters

  • bbox – A list of bounding box

  • labels – Integer to label your bounding box

  • expand – Expand the bounding box to search for neighbors

  • scale – Scale the bounding box to search for neighbors

Returns

A list of neighbors

spatialtis_core.comb_bootstrap(exp_matrix: numpy.ndarray, markers: List[str], neighbors: spatialtis_core.types.Neighbors, labels: spatialtis_core.types.Labels, pval: float = 0.05, times: int = 1000) → List[Tuple[str, str, float]]#

Bootstrap between two types

If you want to test co-localization between protein X and Y, first determine if the cell is X-positive and/or Y-positive. True is considered as positive and will be counted.

Parameters

  • exp_matrix – The expression matrix, each row should be a marker

  • markers – Match to the row of exp_matrix

  • neighbors – List of neighbors

  • labels – List of labels

  • pval – The threshold of p-value

  • times – How many times to perform bootstrap

Returns

The significance between markers List of (marker1, marker2, p-value)

spatialtis_core.dumps_points_wkt(points: spatialtis_core.types.Points) → List[str]#

Dumps points into wkt format

Parameters

points – A list of 2D points

Returns

A list of wkt string

spatialtis_core.dumps_polygons_wkt(polygons: List[spatialtis_core.types.Points]) → List[str]#

Dumps points into wkt format

Parameters

polygons – A list of polygons

Returns

A list of wkt string

spatialtis_core.fast_corr(data1: numpy.ndarray, data2: numpy.ndarray, method: Optional[str] = 'pearson') → numpy.ndarray#

Parallel pairwise correlation

Compute pairwise (combination with replacement) correlation between two 2D array with same shape. Faster than scipy’s implementation.

Parameters

  • data1 – 2D Array

  • data2 – 2D Array

  • method – “pearson” or “spearman”

spatialtis_core.getis_ord(points: spatialtis_core.types.Points, bbox: spatialtis_core.types.BoundingBox, search_level: int = 3, quad: Optional[Tuple[int, int]] = None, rect_side: Optional[Tuple[float, float]] = None, pval: float = 0.05, min_cells: int = 10) → List[bool]#

Getis-ord analysis to find hot cells

Parameters

  • points – A list of points

  • bbox – The bounding box

  • search_level – The level of outer-ring to search for

  • quad – eg.(X, Y) Use X * Y grid to perform analysis

  • rect_side – eg.(X, Y) Use X * Y rectangle to perform analysis

  • pval – The threshold for p-value

  • min_cells – The minimum number of cells to perform analysis

Returns

A list of bool

spatialtis_core.multipoints_bbox(points_collections: Sequence[spatialtis_core.types.Points]) → Sequence[spatialtis_core.types.BoundingBox]#

A utility function to return minimum bounding box list of polygons

Parameters

points_collections – List of 2d points collections

Returns

A list of bounding box (minx, miny, maxx, maxy)

spatialtis_core.multipolygons_area(points_collections: Sequence[spatialtis_core.types.Points]) → Sequence[float]#

Calculate the area of polygons using shoelace formula

Parameters

points_collections – List of 2d points collections, represents a list of polygons

Returns

The area of the polygons in list

spatialtis_core.neighbor_components(neighbors: spatialtis_core.types.Neighbors, labels: spatialtis_core.types.Labels, types: List[str])#

Compute the number of different cells at neighbors

Parameters

  • neighbors – The neighbors dict

  • labels – Integer to label points

  • types – A list of types match to points

Returns

can be used to construct dataframe

Return type

(header, data)

spatialtis_core.points_bbox(points: spatialtis_core.types.Points) → spatialtis_core.types.BoundingBox#

Return minimum bounding box of points

Parameters

points – A list of points

Returns

The bounding box (minx, miny, maxx, maxy) or (minx, miny, minz, maxx, maxy, maxz)

spatialtis_core.points_neighbors(points: spatialtis_core.types.Points, labels: spatialtis_core.types.Labels, r: Optional[float] = None, k: Optional[int] = None, method: str = 'kdtree') → List[List[int]]#

Get neighbors for each points

When search with KD-tree, you can use either r or k. If r = 5, it will search for all points within 5; If k = 5, it will search for the 5 nearest neighbors; If combined, r = 5 and k = 5, it will search for at most 5 neighbors within 5;

When search with Delaunay triangulation, there is no parameter.

The return list follow the order of labels. For example, if serach for points of [100, 101, 102], the result may look like [[100, 102], [101, 102], [102, 100, 101]]

Parameters

  • points – A list of points

  • labels – Integer to labels your points

  • r – Radius range to search for neighbors

  • k – Number of nearest neighbors

  • method – “kdtree” or “delaunay”

Returns

A list of neighbors

spatialtis_core.points_shapes(polygons: List[spatialtis_core.types.Points], method: str = 'convex', concavity: float = 1.5) → spatialtis_core.types.Points#

Acquire multipoints (shapes) that describe the points

Parameters

  • polygons – A list of polygons

  • method – “convex” or “concave”

  • concavity – Determine the concavity in concave hull

Returns

A list of points

spatialtis_core.polygons_area(points: spatialtis_core.types.Points) → float#

Calculate the area of polygons using shoelace formula

Parameters

points – A list of points represents a polygon

Returns

The area of the polygon

spatialtis_core.reads_wkt_points(wkt_strings: List[str]) → spatialtis_core.types.Points#

Reads wkt points into python object

Parameters

wkt_strings – A list of wkt string represents points

Returns

A list of 2D points

spatialtis_core.reads_wkt_polygons(wkt_strings: List[str]) → List[spatialtis_core.types.Points]#

Reads wkt points into python object

Parameters

wkt_strings – A list of wkt string represents polygons

Returns

A list of 2D polygons

spatialtis_core.spatial_autocorr(x: numpy.ndarray, neighbors: spatialtis_core.types.Neighbors, labels: spatialtis_core.types.Labels, two_tailed: bool = True, pval: float = 0.05, method: str = 'moran_i') → List[Tuple[float, float]]#

Compute spatial auto-correlation value for a 2D array in parallel

The p-value is under the assumption of normal distribution Return is tuples of (spatial_autocorr value, p value)

Parameters

  • x – Gene expression matrix, each row is the expression of a gene

  • neighbors – A list of neighbors

  • labels – A list of labels

  • two_tailed – Determine the p value

  • pval – The p-value threshold

  • method – “moran_i” or “geary_c”

Returns

A list of (value, p_value)

spatialtis_core.spatial_distribution_pattern(points_collections: List[spatialtis_core.types.Points], bbox: spatialtis_core.types.BoundingBox, method: str = 'id', r: Optional[float] = None, resample: int = 1000, quad: Optional[Tuple[int, int]] = None, rect_side: Optional[Tuple[float, float]] = None, pval: float = 0.05, min_cells: int = 10, dims: int = 2) → List[Tuple[float, float, int]]#

Compute the distribution index and determine the pattern for different cells in a ROI in parallel

If data is 3D, only method=”id” is supported

Parameters

  • points_collections – A list of list of points

  • bbox – The bounding box

  • method – “id” for index of dispersion, “morisita” for morisita index, “clark_evans” for clark evans’ index

  • r – If method == “id”; The sample windows’ radius

  • resample – If method == “id”; The number of sampling times

  • quad – If method == “morisita”; eg.(X, Y) Use X * Y grid to perform analysis

  • rect_side – If method == “morisita”; eg.(X, Y) Use X * Y rectangle to perform analysis

  • pval – The threshold for p-value

  • min_cells – The minimum number of cells to perform analysis

  • dims – The dimension of data

Returns

A list of (index_value, p_value, pattern)

spatialtis_core.spatial_entropy(points_collections: List[spatialtis_core.types.Points], types_collections: List[List[str]], method: str = 'leibovici', d: Optional[float] = None, cut: int = 3, dims: int = 2) → List[float]#

Compute spatial entropy value of multiple ROIs in parallel

Parameters

  • points_collections – A list of list of points

  • types_collections – A list of list of types

  • bbox – The bounding box

  • method – “leibovici” or “altieri”

  • d – If method == “leibovici”; The distance threshold to determine co-occurrence

  • cut – If method == “altieri”; The distance interval to determine co-occurrence

  • dims – The dimension of data

Returns

A list of spatial entropy

spatialtis_core.spatial_weight(neighbors: types.Neighbors, labels: types.Labels) → scipy.sparse.csr_matrix#

Build a neighbors sparse matrix from neighbors data

Parameters

  • neighbors – List of neighbors

  • labels – List of neighbors

Returns

A scipy sparse matrix in csr format

spatialtis_core.Labels#
spatialtis_core.Neighbors#