definition.yml

method:
  id: paga_tree
  name: PAGA Tree
  tool_id: paga
  source: tool
  platform: Python
  url: https://github.com/theislab/graph_abstraction
  authors:
  - given: Alexander
    family: Wolf
    email: alex.wolf@helmholtz-muenchen.de
    github: falexwolf
    ORCID: 0000-0002-8760-7838
  - given: Fabian
    family: Theis
    email: fabian.theis@helmholtz-muenchen.de
    github: theislab
  description: >
    This method runs exactly the same as normal PAGA, but will construct a minimal-spanning tree between clusters

wrapper:
  type: branch_trajectory
  topology_inference: free
  trajectory_types:
  - linear
  - bifurcation
  - convergence
  - multifurcation
  - tree
  input_required:
  - counts
  - start_id
  input_optional:
  - groups_id

container:
  docker: dynverse/ti_paga_tree
  url: https://github.com/dynverse/ti_paga_tree

manuscript:
  doi: 10.1186/s13059-019-1663-x
  google_scholar_cluster_id: '10470081259069082868'
  preprint_date: '2017-10-27'
  publication_date: '2019-03-19'

parameters:
- id: filter_features
  description: Whether to do feature filtering
  type: logical
  default: yes
- id: n_neighbors
  description: Number of neighbours for knn
  type: integer
  default: 15
  distribution:
    type: uniform
    lower: 1
    upper: 100
- id: n_comps
  description: Number of principal components
  type: integer
  default: 50
  distribution:
    type: uniform
    lower: 0
    upper: 100
- id: n_dcs
  description: Number of diffusion components for denoising graph, 0 means no denoising.
  type: integer
  default: 15
  distribution:
    type: uniform
    lower: 0
    upper: 40
- id: resolution
  description: Resolution of louvain clustering, which determines the granularity
    of the clustering. Higher values will result in more clusters.
  type: numeric
  default: 1
  distribution:
    type: uniform
    lower: 0.1
    upper: 10
- id: embedding_type
  description: Either 'umap' (scales very well, recommended for very large datasets)
    or 'fa' (ForceAtlas2, often a bit more intuitive for small datasets).
  type: character
  default: fa
  values:
  - umap
  - fa