Update docs for using growth rate with OT

docs/source/getting_started.rst

@@ -99,7 +99,7 @@ If the dataset has only one clonal time point, you can run::
 
     adata=cs.tmap.infer_Tmap_from_one_time_clones(adata_orig,initial_time_points=None, later_time_point=None,initialize_method='OT',**params)
 
-which jointly optimizes the transition map and the initial clonal structure. It requires initializing the transition map using state information alone. We provide two methods for such initialization: 1) ``OT`` for using the standard optimal transport approach; 2) ``HighVar`` for a customized approach, assuming that cells similar in gene expression across time points share clonal origin. Depending on the choice,  the initialized map is stored at ``adata.uns['OT_transition_map']`` or  ``adata.uns['HighVar_transition_map']``. The final product is stored at ``adata.uns['transition_map']``.
+which jointly optimizes the transition map and the initial clonal structure. It requires initializing the transition map using state information alone. We provide two methods for such initialization: 1) ``OT`` for using the standard optimal transport approach (if you wish to utilize the growth rate information as Waddington-OT, you can directly pass the growth rate estimate for each cell to the input AnnaData object `adata_orig.obs["cell_growth_rate"]`); 2) ``HighVar`` for a customized approach, assuming that cells similar in gene expression across time points share clonal origin. Depending on the choice,  the initialized map is stored at ``adata.uns['OT_transition_map']`` or  ``adata.uns['HighVar_transition_map']``. The final product is stored at ``adata.uns['transition_map']``.
 
 ``HighVar`` converts highly variable genes into pseudo multi-time clones and infers a putative map with coherent sparse optimization. We find the `HighVar` method performs better than the `OT` method, especially when there are large differentiation effects over the observed time window, or batch effects.