Faster RTD & fix CI

.readthedocs.yaml

@@ -1,17 +1,18 @@
-# https://docs.readthedocs.io/en/stable/config-file/v2.html
 version: 2
 build:
-  os: ubuntu-20.04
+  os: ubuntu-24.04
   tools:
-    python: "3.10"
-sphinx:
-  configuration: docs/conf.py
-  # disable this for more lenient docs builds
-  fail_on_warning: true
-python:
-  install:
-    - method: pip
-      path: .
-      extra_requirements:
-        - doc
-        - full
+    python: "3.13"
+  jobs:
+    create_environment:
+      - asdf plugin add uv
+      - asdf install uv latest
+      - asdf global uv latest
+      - uv venv
+      - uv pip install .[doc,full]
+    build:
+      html:
+        - uv run sphinx-build -T -W -b html docs $READTHEDOCS_OUTPUT/html
+
+submodules:
+    include: all

CHANGELOG.md

@@ -27,7 +27,7 @@ All functions have been rewritten to follow the new API, errors when running pre
   - `viper` now correctly estimates shadow regulons when network weights are values other than -1 or +1
   - `wsum` and `wmean` are deprecated, instead now the method `waggr` allows to run both methods and any custom function. This makes it easier to quickly test new enrichment methods without having to deal with `decoupler`'s implementation
 - Databases from Omnipath can now be accessed through the new `op` module
-  - Use `decoupler.op.<resource_name>` to access a database  
+  - Use `decoupler.op.<resource_name>` to access a database
   - Removed the `omnipath` package as a dependancy
   - Fixed `collectri` to the publication version instead of the OmniPath one
   - Made `progeny` only return significant genes by default instead of the top N genes per pathway
@@ -63,7 +63,7 @@ All functions have been rewritten to follow the new API, errors when running pre
   - Added `obsbar` to plot size of metadata columns in `anndata.AnnData.obs`
   - Added `order` to plot sources or features along a continous process such as a trajectory
   - Added `order_targets` to plot the targets of a given source along a continous process
-  - 
+  -
 - New preprocessing functions in the `pp` module
   - Added two functions to format networks, `adjmat` to return an adjacency matrix, and `idxmax` to return a list of sets
   - Added `filter_samples` to filter pseudobulk profiles after running `pseudobulk`

README.md

@@ -1,6 +1,6 @@
 # decoupler - Ensemble of methods to infer enrichment scores
 <img src="https://github.com/saezlab/decoupleR/blob/master/inst/figures/logo.svg?raw=1" align="right" width="120" class="no-scaled-link" />
-   
+
 
 [![Tests][badge-tests]][tests]
 [![Documentation][badge-docs]][documentation]

docs/_static/css/custom.css

@@ -5,10 +5,10 @@ div.cell_output table.dataframe {
 
 /* Adjust the logo size */
 .logo img {
-    width: 50%;   /* or any percentage you want */
+    width: 50%; /* or any percentage you want */
     height: auto; /* maintain aspect ratio */
 }
 
 img.no-scaled-link {
     background: transparent !important;
-}
+}

docs/api/index.md

@@ -21,4 +21,4 @@ pp
 
 pl
 tl
-```
+```

docs/api/mt.md

@@ -30,4 +30,4 @@
 
     mt.decouple
     mt.consensus
-```
+```

docs/api/op.md

@@ -25,4 +25,4 @@
     op.show_resources
     op.show_organisms
     op.translate
-```
+```

docs/api/pl.md

@@ -20,4 +20,4 @@
     pl.order
     pl.source_targets
     pl.volcano
-```
+```

docs/references.bib

@@ -35,7 +35,7 @@ @article{pydeseq
   author = {Muzellec,  Boris and Teleńczuk,  Maria and Cabeli,  Vincent and Andreux,  Mathieu},
   editor = {Ponty,  Yann},
   year = {2023},
-  month = sep 
+  month = sep
 }
 
 @article{limma,
@@ -80,7 +80,7 @@ @article{decoupler
   author = {Badia-i-Mompel,  Pau and Vélez Santiago,  Jesús and Braunger,  Jana and Geiss,  Celina and Dimitrov,  Daniel and M\"{u}ller-Dott,  Sophia and Taus,  Petr and Dugourd,  Aurelien and Holland,  Christian H and Ramirez Flores,  Ricardo O and Saez-Rodriguez,  Julio},
   editor = {Kuijjer,  Marieke Lydia},
   year = {2022},
-  month = jan 
+  month = jan
 }
 
 @article{grn_review,
@@ -210,7 +210,7 @@ @article{panglaodb
   publisher = {Oxford University Press (OUP)},
   author = {Franzén,  Oscar and Gan,  Li-Ming and Bj\"{o}rkegren,  Johan L M},
   year = {2019},
-  month = jan 
+  month = jan
 }
 
 @article{msvisium,
@@ -254,7 +254,7 @@ @article{hcop
   publisher = {Oxford University Press (OUP)},
   author = {Yates,  Bethan and Gray,  Kristian A and Jones,  Tamsin E M and Bruford,  Elspeth A},
   year = {2021},
-  month = may 
+  month = may
 }
 
 @article{omnipath,
@@ -325,7 +325,7 @@ @article{mofacell
   publisher = {eLife Sciences Publications,  Ltd},
   author = {Ramirez Flores,  Ricardo Omar and Lanzer,  Jan David and Dimitrov,  Daniel and Velten,  Britta and Saez-Rodriguez,  Julio},
   year = {2023},
-  month = nov 
+  month = nov
 }
 
 @article{scanpy,
@@ -339,7 +339,7 @@ @article{scanpy
   publisher = {Springer Science and Business Media LLC},
   author = {Wolf,  F. Alexander and Angerer,  Philipp and Theis,  Fabian J.},
   year = {2018},
-  month = feb 
+  month = feb
 }
 
 @article{paga,
@@ -353,7 +353,7 @@ @article{paga
   publisher = {Springer Science and Business Media LLC},
   author = {Wolf,  F. Alexander and Hamey,  Fiona K. and Plass,  Mireya and Solana,  Jordi and Dahlin,  Joakim S. and G\"{o}ttgens,  Berthold and Rajewsky,  Nikolaus and Simon,  Lukas and Theis,  Fabian J.},
   year = {2019},
-  month = mar 
+  month = mar
 }
 
 @article{pax5,
@@ -412,7 +412,7 @@ @article{gsva
   publisher = {Springer Science and Business Media LLC},
   author = {H\"{a}nzelmann,  Sonja and Castelo,  Robert and Guinney,  Justin},
   year = {2013},
-  month = jan 
+  month = jan
 }
 
 @inproceedings{xgboost,
@@ -470,5 +470,5 @@ @article{zscore
   publisher = {Springer Science and Business Media LLC},
   author = {Yılmaz,  Serhan and Ayati,  Marzieh and Schlatzer,  Daniela and undefinedi\c{c}ek,  A. Erc\"{u}ment and Chance,  Mark R. and Koyut\"{u}rk,  Mehmet},
   year = {2021},
-  month = feb 
+  month = feb
 }

pyproject.toml

@@ -23,21 +23,16 @@ classifiers = [
   "Programming Language :: Python :: 3.13",
 ]
 dependencies = [
+  "adjusttext",
   "anndata",
-  "numba",
   "docrep",
-  "tqdm",
-  "requests",
   "marsilea",
-  "adjustText",
+  "numba",
+  "requests",
+  "scipy<1.16", # https://github.com/statsmodels/statsmodels/issues/9584
   # for debug logging (referenced from the issue template)
   "session-info2",
-]
-optional-dependencies.full = [
-  "igraph",
-  "xgboost",
-  "ipywidgets",
-  "dcor",
+  "tqdm",
 ]
 optional-dependencies.dev = [
   "pre-commit",
@@ -59,16 +54,22 @@ optional-dependencies.doc = [
   "sphinxcontrib-bibtex>=1",
   "sphinxext-opengraph",
 ]
+optional-dependencies.full = [
+  "dcor",
+  "igraph",
+  "ipywidgets",
+  "xgboost",
+]
 optional-dependencies.test = [
   "coverage",
-  "pytest",
-  "statsmodels",
+  "dcor",
   "gseapy",
-  "scanpy",
   "igraph",
-  "xgboost",
   "ipywidgets",
-  "dcor"
+  "pytest",
+  "scanpy",
+  "statsmodels",
+  "xgboost",
 ]
 # https://docs.pypi.org/project_metadata/#project-urls
 urls.Documentation = "https://decoupler.readthedocs.io/"

src/decoupler/bm/metric/_hmean.py

@@ -79,4 +79,4 @@ def hmean(
     hdf['score'] = hdf[h_cols].mean(axis=1, numeric_only=True)
     hdf['score'] = (hdf['score'] - hdf['score'].min()) / (hdf['score'].max() - hdf['score'].min())
     return hdf
-        
+

src/decoupler/mt/_aucell.py

@@ -71,13 +71,13 @@ def _func_aucell(
     the set appear in this ranking. Specifically, the enrichment score :math:`ES` is:
 
     .. math::
-    
+
        {ES}_{i, F} = \int_0^1 {RecoveryCurve}_{i, F}(r_i) \, dr
-    
+
     Where:
-    
-    - :math:`i` is the obervation  
-    - :math:`F` is the feature set  
+
+    - :math:`i` is the obervation
+    - :math:`F` is the feature set
     - :math:`{RecoveryCurve}_{i, F}(r_i)` is the proportion of features from :math:`F` recovered in the top :math:`r_i`-fraction of the ranked list for observation :math:`i`
 
     %(notest)s
@@ -92,7 +92,7 @@ def _func_aucell(
     nobs, nvar = mat.shape
     nsrc = starts.size
     n_up = _validate_n_up(nvar, n_up)
-    m = f'aucell - calculating {nsrc} AUCs for {nvar} targets across {nobs} observations, categorizing features at rank={n_up}' 
+    m = f'aucell - calculating {nsrc} AUCs for {nvar} targets across {nobs} observations, categorizing features at rank={n_up}'
     _log(m, level='info', verbose=verbose)
     es = np.zeros(shape=(nobs, nsrc))
     for i in tqdm(range(mat.shape[0]), disable=not verbose):

src/decoupler/mt/_consensus.py

@@ -70,7 +70,7 @@ def consensus(
 
     For each method, enrichment scores are split into positive and negative subsets
     and transformed independently into z-scores.
-    
+
     1. Subset values based on sign (direction).
     2. Mirror each subset into positive and negative values with the same magnitude.
     3. Compute z-scores for each subset: :math:`z_i = \frac{x_i - \mu}{\sigma}`.
@@ -83,7 +83,7 @@ def consensus(
 
     .. math::
 
-        ES = \frac{\sum_{m=1}^{M} z_{i}^{(m)}}{M} 
+        ES = \frac{\sum_{m=1}^{M} z_{i}^{(m)}}{M}
 
     Where:
 

src/decoupler/mt/_gsea.py

@@ -169,8 +169,8 @@ def _func_gsea(
     statistic when a feature is in the set, and decreasing it when it is not.
 
     .. math::
-    
-       \delta(F, i) = 
+
+       \delta(F, i) =
        \begin{cases}
        \frac{|r_i|}{\sum\limits_{j \in F} |r_j|} & \text{if feature } i \in F \\
        -\frac{1}{l} & \text{if feature } i \notin F

src/decoupler/mt/_gsva.py

@@ -85,7 +85,7 @@ def _mat_ecdf(
     for j in range(mat.shape[1]):
         D[:, j] = _ecdf(mat[:, j])
     return D
-    
+
 
 @nb.njit(cache=True)
 def _col_d(
@@ -317,19 +317,19 @@ def _func_gsva(
     Gene Set Variation Analysis (GSVA) :cite:`gsva`.
 
     Each feature is first transformed and smoothed using a kernel density estimation method:
-    
+
     - Gaussian
     - Poisson
     - Empirical cumulative distribution function
 
     Features are then ranked based on a continuous metric (e.g., expression value, score, or correlation).
-    
+
     Then, a score for each feature in a set is computed by walking down the ranked list,
     increasing a running-sum statistic when a feature belongs to the set and decreasing it otherwise.
 
     .. math::
-    
-       \delta(F, i) = 
+
+       \delta(F, i) =
        \begin{cases}
        \frac{|r_i|}{\sum\limits_{j \in F} |r_j|} & \text{if feature } i \in F \\
        -\frac{1}{l} & \text{if feature } i \notin F

src/decoupler/mt/_mdt.py

@@ -41,7 +41,7 @@ def _func_mdt(
     which are the weights of all feature sets :math:`F`. It uses the implementation provided by ``xgboost`` :cite:`xgboost`.
 
     The enrichment score :math:`ES` for each :math:`F` is then calculated as the importance of each covariate in the model.
-    
+
     %(notest)s
 
     %(params)s

src/decoupler/mt/_mlm.py

@@ -87,7 +87,7 @@ def _func_mlm(
     .. math::
 
         p_{value} = 2 \times \mathrm{sf}(|ES|, \text{df})
-    
+
 
     %(params)s
     %(tval)s

src/decoupler/mt/_ora.py

@@ -179,11 +179,11 @@ def _func_ora(
        :width: 100%
 
        Over Representation Analysis (ORA) scheme.
-    
+
     The statistic is calculated as the Odds Ratio :math:`OR` with Haldane-Anscombe correction.
 
     .. math::
-        
+
         \text{OR} = \log{\frac{\frac{a + 0.5}{b + 0.5}}{\frac{c + 0.5}{d + 0.5}}}
 
     And the :math:`p_{value}` is obtained afer computing a two-tailed Fisher’s exact test with the same table.
@@ -206,16 +206,16 @@ def _func_ora(
     nsrc = starts.size
     if n_up is None:
         n_up = int(np.max([np.ceil(0.05 * nvar), 2]))
-        m = f'ora - setting n_up={n_up}' 
+        m = f'ora - setting n_up={n_up}'
         _log(m, level='info', verbose=verbose)
     if n_bg is None:
         n_bg = 0
-        m = f'ora - not using n_bg, a feature specific background will be used instead' 
+        m = f'ora - not using n_bg, a feature specific background will be used instead'
         _log(m, level='info', verbose=verbose)
     assert isinstance(n_up, (int, float)) and n_up > 0, 'n_up must be numeric and > 0'
     assert isinstance(n_bm, (int, float)) and n_bm >= 0, 'n_bm must be numeric and positive'
     assert isinstance(n_bg, (int, float)) and n_bg >= 0, 'n_bg must be numeric and positive'
-    m = f'ora - calculating {nsrc} scores across {nobs} observations with n_up={n_up}, n_bm={n_bm}, n_bg={n_bg}' 
+    m = f'ora - calculating {nsrc} scores across {nobs} observations with n_up={n_up}, n_bm={n_bm}, n_bg={n_bg}'
     _log(m, level='info', verbose=verbose)
     es = np.zeros((nobs, nsrc))
     pv = np.zeros((nobs, nsrc))

src/decoupler/mt/_udt.py

@@ -57,7 +57,7 @@ def _func_udt(
     _check_import(xgboost, 'xgboost')
     nobs = mat.shape[0]
     nvar, nsrc = adj.shape
-    m = f'udt - fitting {nsrc} univariate decision tree models (XGBoost) of {nvar} targets across {nobs} observations' 
+    m = f'udt - fitting {nsrc} univariate decision tree models (XGBoost) of {nvar} targets across {nobs} observations'
     _log(m, level='info', verbose=verbose)
     es = np.zeros(shape=(nobs, nsrc))
     for i in tqdm(range(nobs), disable=not verbose):