Update README.md

README.md

@@ -5,19 +5,23 @@ The AIDO.Tissue model is develpoed on spatial single-cell transcriptomic data. T
 
 We evelauted the model on two spatial data task, including predicting niche label and cell density. The metrics are as below:
 
-| Task    | F1-score |
+niche label type prediction:
+| Model    | F1-score |
 | -------- | ------- |
-| niche label type prediction  | 0.67    |
+|  AIDO.Tissue | 0.67    |
+|  Nicheformer | 0.50    |
 
+cell density prediction:
 | Task    | Mean absolute error | R square |
 | -------- | ------- | ------- |
-| cell density prediction  | 4.44  | 0.55|
+|  AIDO.Tissue | 4.44  | 0.55|
+|  Nicheformer | 7.08  | -0.07|
 
 # Finetuning AIDO.Tissue for spatial single cell downstream tasks
 We introduce how to finetune and evaluate our pre-trained AIDO.Tissue foundation models for downstream tasks. These tasks can be classified into the following categories:
 
-  * **Sequence-level classification tasks**: niche label type prediction
-  * **Sequence-level regression tasks**: cell density prediction
+  * **Cell-level classification tasks**: niche label type prediction
+  * **Cell-level regression tasks**: cell density prediction
 
 Note: All the following scripts should be run under `ModelGenerator/`.
 
@@ -28,7 +32,7 @@ For each `.h5ad`, several obs attributes should be included to reprezent the spa
 
 Note: the file `scRNA_genename_and_index.tsv` includes all the corresponding gene name and index in h5ad file.
 
-## Sequence-level classification tasks
+## Cell-level classification tasks
 ### niche label type prediction
 We fully finetune AIDO.Tissue for niche label type prediction.
 
@@ -56,7 +60,7 @@ CUDA_VISIBLE_DEVICES=6 nohup mgen test --config experiments/AIDO.Tissue/niche_ty
 Note: `ckpt_path` is the finetuned checkpoint path.
 
 
-## Sequence-level regression tasks
+## Cell-level regression tasks
 
 ### cell density prediction