Update Readme with scRNA information

README.md

@@ -28,96 +28,42 @@ This repository provides the HR-VILAGE-3K3M dataset, a curated collection of hum
 **Fig**: Overview of HV-RIGEL-3K. (a) HV-RIGEL-3K construction workflow. (b) Distribution of sample timepoints for vaccine and inoculation studies. (c) Composition of the dataset by platform, tissue type, study type, and pathogen.
 
 # Dataset Description:
-- This repo contains 59 studies, comprising 2393 subjects and 13,515 observations.
-- We provide preprocessed and normalized gene expression data, raw gene expression data, meta data and antibody data.
+- This repo contains 66 studies (59 bulk and 7 single cell studies), comprising 3178 subjects, 14,136 observations, and 2,557,942 single cells.
+- We provide preprocessed and normalized gene expression data, raw gene expression data, metadata and antibody data.
 
 # Data Structure:
 ```
 HR-VILAGE-3K3M/
 ├── README.md
 ├── study_meta.csv
-├── gene_expr/
+├── bulk_gene_expr/
 │    └── <study_ID>_gene_expr.csv
+├── single_cell_gene_expr/
+│    └── <study_ID>_processed.h5ad
 ├── meta/
 │    └── <study_ID>_meta.csv
-├── gene_expr_raw/
+├── bulk_gene_expr_raw/
 │    └── <study_ID>_raw.csv
 └── antibody/
      └── <study_ID>_antibody.csv
 ```
 - **study_meta.csv**: Contains study-level metadata (e.g., platform, tissue type, study type) and serves as an overview of the repository. Users can use this file to filter and select a subset of studies based on custom criteria.
-- **gene_expr/**: Processed gene expression matrices for each study (sample-by-gene). All files share the same 41,667 gene columns, with 9,004 genes nonmissing across all studies.
+- **bulk_gene_expr/**: Processed gene expression matrices for each study (sample-by-gene). All files share the same 41,667 gene columns, with 9,004 genes non-missing across all studies.
+- **single_cell_gene_expr/**: Contains processed .h5ad files for each single-cell RNA-seq study. Raw count matrices are stored in the .X attribute, and cell-level metadata is stored in the .obs dataframe. 
 - **meta/**: Study-specific metadata files (.csv), aligned by row names with the corresponding expression data. All metadata files share the same column structure; missing values are marked as NA.
-- **gene_expr_raw/**: Raw probe-level expression matrices (probe-by-sample), provided when available to support custom preprocessing workflows.
+- **bulk_gene_expr_raw/**: Raw probe-level expression matrices (probe-by-sample), provided when available to support custom preprocessing workflows.
 - **antibody/**: Raw antibody measurements with sample IDs matching those in the metadata, enabling integration with gene expression data at the subject level.
 
-# Data Relations:
-The following duplicate studies (right nodes) are not included in HR-VILAGE-3K3M, while their source studies (left leaves) are included in HR-VILAGE-3K3M.
-```
-GSE73072_H3N2_DEE2 ──┐
-                      ├── GSE52428
-GSE73072_H1N1_DEE4 ──┘
-
-GSE73072_H3N2_DEE2    ├── GSE30550
-
-GSE73072_HRV_UVA   ──┐
-GSE73072_RSV_DEE1  ── ├── GSE17156
-GSE73072_H3N2_DEE2 ──┘
-```
-
-The studies within the following groups are from the same study group with less batch effect.
-- **Group 1**
-  | E_MTAB_12993_blood | E_MTAB_12993_nose |
-  |---|---|
-- **Group 2**
-  | GSE61754 | GSE90732 |
-  |---|---|
-- **Group 3**
-  | GSE73072_H1N1_DEE4 | GSE73072_H1N1_DEE3 | GSE73072_H3N2_DEE2 | GSE73072_H3N2_DEE5 | GSE73072_HRV_DUKE | GSE73072_HRV_UVA | GSE73072_RSV_DEE1 |
-  |---|---|---|---|---|---|---|
-- **Group 4**
-  | GSE74811 | GSE74813 | GSE74815 | GSE74816 |
-  |---|---|---|---|
-- **Group 5**
-  | GSE48023 | GSE48018 |
-  |---|---|
-- **Group 6**
-  | GSE59635 | GSE59654 |
-  |---|---|
-- **Group 7**
-  | GSE59743 | GSE101709 | GSE101710 |
-  |---|---|---|
-- **Group 8**
-  | GSE190001_PRIME | GSE190001_BOOST |
-  |---|---|
-- **Group 9**
-  | SDY311 | SDY112 | SDY315 |
-  |---|---|---|
-- **Group 10**
-  | GSE201533 | GSE190747 | GSE201642 |
-  |---|---|---|
-- **Group 11**
-  | GSE246525 | GSE276544 | GSE247401 |
-  |---|---|---|
-- **Group 12**
-  | GSE169159 | GSE102012 |
-  |---|---|
-- **Group 13**
-  | GSE29614 | GSE29615 | GSE29617 |
-  |---|---|---|
-- **Group 14**
-  | GSE52005 | GSE48762 |
-  |---|---|
-
 # How to start:
-Users could directly download all files and read the files locally.
-Alternaitvely, the following provide (partially) loading the dataset into Python using `dataset` package.
+Users could directly download all files and read files locally.
+Alternatively, the following provides (partially) loading the dataset into Python using `dataset` package.
 ```python
 repo_id = "xuejun72/HR-VILAGE-3K3M"
 import pandas as pd
 from datasets import load_dataset
 ```
-There are two ways to load (and combine) the datasets.
+## Bulk gene expression data
+Bulk gene expression data can be loaded and combined using two alternative approaches.
 1. Use our predefined configuration name, and pass to `name` argument in `load_dataset()`. `trust_remote_code=True` is required.
 
    Example 1, to download study_meta:
@@ -131,13 +77,13 @@ There are two ways to load (and combine) the datasets.
    ```
    Example 3, to download and combine **all** gene expression datasets. However, this is highly **NOT** recommended since their size are too large and the execution time will be long.
    ```python
-   # Not recommened!
+   # Not recommended!
    gene_expr_dict = load_dataset(repo_id, name = "gene_expr", trust_remote_code=True)
    gene_expr = gene_expr_dict["train"].to_pandas()
    ```
 
-   In addition, we provide study filter function before downloading and loading, which works for **meta** and **gene expression** datasets.
-   `split_filter` argument is designed for this filter, which is optional. By default, `split_filter=None` will downloading all datasets as shown before.
+   In addition, we provide a study filter function before downloading and loading, which works for **meta** and **bulk_gene_expr** datasets.
+   `split_filter` argument is designed for this filter, which is optional. By default, `split_filter=None` will download all datasets as shown before.
    `split_filter` is a `dict` Python object where `key` is filter factors taking values from `['study_type','platform','tissue','pathogen','vaccine']` and `value` is a list of categories for each `key`. `value` should be exact same as that in study_meta.csv.
    Some examples of a valid `split_filter`:
    ```python
@@ -145,7 +91,7 @@ There are two ways to load (and combine) the datasets.
    split_filter = {"study_type": ["vaccine"], "vaccine": ["Influenza TIV"]}
    split_filter = {"study_type": ["vaccine"], "platform": ["RNA-seq"], "tissue": ["PBMC","nasal swab"], "pathogen": []}
    ```
-   Example 4, to download and combine a customized filtered meta datasets:
+   Example 4, to download and combine a customized filtered meta dataset:
    ```python
    split_filter = {"study_type": ["vaccine"], "platform": ["RNA-seq"], "tissue": ["PBMC","nasal swab"], "pathogen": []}
    meta_filtered_dict = load_dataset(repo_id, name = "meta", trust_remote_code=True, split_filter=split_filter)
@@ -153,7 +99,7 @@ There are two ways to load (and combine) the datasets.
      meta_filtered = value.to_pandas().set_index("row_name")
    meta_filtered
    ```
-   Example 5, to download and combine a customized filtered gene expression datasets:
+   Example 5, to download and combine a customized filtered gene expression dataset:
    ```python
    split_filter = {"study_type": ["vaccine"], "platform": ["RNA-seq"], "tissue": ["PBMC","nasal swab"], "pathogen": []}
    gene_expr_filtered_dict = load_dataset(repo_id, name = "gene_expr", trust_remote_code=True, split_filter=split_filter)
@@ -161,7 +107,7 @@ There are two ways to load (and combine) the datasets.
      gene_expr_filtered = value.to_pandas().set_index("row_name")
    gene_expr_filtered
    ```
-3. Use exact path of one csv file, and pass to `data_files` argument in `load_dataset()`.
+2. Use the exact path of one csv file, and pass to `data_files` argument in `load_dataset()`.
 
    Example 1, to download study_meta:
    ```python
@@ -177,8 +123,95 @@ There are two ways to load (and combine) the datasets.
    ```
    Note: for **antibody** and **raw gene expression** datasets, since different study has different columns which cannot be simply combined, loading them must using `data_files` argument and be one-by-one.
 
+## Single cell gene expression data
+Single-cell gene expression data can be downloaded and accessed using the anndata package.
+
+  ```python
+    import anndata as ad
+    # Load the GSE195673 dataset
+    GSE195673 = ad.read_h5ad("./GSE195673_processed.h5ad")
+    # View cell-level metadata
+    GSE195673.obs
+  ```
+A merged dataset containing all 7 studies is also provided, comprising 2,557,942 cells and 13,589 common genes:
+  ```python
+    import anndata as ad
+    # Load the combined dataset
+    combined = ad.read_h5ad("./combined.h5ad")
+    # View cell-level metadata
+    combined.obs
+  ```
+The combined object includes detailed cell-level metadata such as sample_id, cell_type, sex, donor_id, time_point_day, dataset, covid_status, age, tissue, and study_type. It also contains dimensionality reductions (X_pca, X_umap) and graph-based neighbor information in obsp for downstream analysis.
+
 # Example code:
 
 
+Single Cell Visualization
+  ```python
+  import scanpy as sc
+  # Visualize UMAP colored by dataset
+  sc.pl.umap(combined, color='dataset', save='_combined_by_dataset.png', show=True)
+  ```
+
+
+# Data Relations:
+The following duplicate studies (right nodes) are not included in HR-VILAGE-3K3M, while their source studies (left leaves) are included in HR-VILAGE-3K3M.
+```
+GSE73072_H3N2_DEE2 ──┐
+                      ├── GSE52428
+GSE73072_H1N1_DEE4 ──┘
+
+GSE73072_H3N2_DEE2    ├── GSE30550
+
+GSE73072_HRV_UVA   ──┐
+GSE73072_RSV_DEE1  ── ├── GSE17156
+GSE73072_H3N2_DEE2 ──┘
+```
+
+The studies within the following groups are from the same study group with less batch effect.
+- **Group 1**
+  | E_MTAB_12993_blood | E_MTAB_12993_nose |
+  |---|---|
+- **Group 2**
+  | GSE61754 | GSE90732 |
+  |---|---|
+- **Group 3**
+  | GSE73072_H1N1_DEE4 | GSE73072_H1N1_DEE3 | GSE73072_H3N2_DEE2 | GSE73072_H3N2_DEE5 | GSE73072_HRV_DUKE | GSE73072_HRV_UVA | GSE73072_RSV_DEE1 |
+  |---|---|---|---|---|---|---|
+- **Group 4**
+  | GSE74811 | GSE74813 | GSE74815 | GSE74816 |
+  |---|---|---|---|
+- **Group 5**
+  | GSE48023 | GSE48018 |
+  |---|---|
+- **Group 6**
+  | GSE59635 | GSE59654 |
+  |---|---|
+- **Group 7**
+  | GSE59743 | GSE101709 | GSE101710 |
+  |---|---|---|
+- **Group 8**
+  | GSE190001_PRIME | GSE190001_BOOST |
+  |---|---|
+- **Group 9**
+  | SDY311 | SDY112 | SDY315 |
+  |---|---|---|
+- **Group 10**
+  | GSE201533 | GSE190747 | GSE201642 |
+  |---|---|---|
+- **Group 11**
+  | GSE246525 | GSE276544 | GSE247401 |
+  |---|---|---|
+- **Group 12**
+  | GSE169159 | GSE102012 |
+  |---|---|
+- **Group 13**
+  | GSE29614 | GSE29615 | GSE29617 |
+  |---|---|---|
+- **Group 14**
+  | GSE52005 | GSE48762 |
+  |---|---|
+
+
 # How to cite: