Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://holab-hku.github.io/starmapVR


https://holab-hku.github.io/starmapVR

Last synced: 30 days ago
JSON representation

Awesome Lists containing this project

README 

        
# StarmapVis: animated visual display of single cell and spatial omic data

Authors: Xiunan Fang, Shichao Ma, Yu Yao, Andrian Yang and Joshua W. K. Ho



Contact: Joshua Ho ([email protected])



## Synopsis



starmap

StarmapVis enables interactive web-based visualisation for various single cell or spatial omic data, including three-dimensional scatter plots with cell type annotation, cellular developmental trajectory, mapping between spatial transcripomics and single cell omic data, and multi-sample comparison views. The intuitive user interface is designed to enhance user experience, especially for producing images or animation for scientific presentation. StarmapVis can be accessed from a desktop, laptop or a mobile device from the following link: [https://holab-hku.github.io/starmapVis/](https://holab-hku.github.io/starmapVis/), or the QR code. 

## Input data



The upload page includes five options to let users upload different input files including:



1. The cell information csv file needs to contain a header row with the following column names – x, y, z and labels – corresponding to the 3D coordinates of points and the cell label (e.g. clusters, gene markers, pseudotime, etc.). 



2. The trajectory information csv file contains the following column names – name, x, y, z, children and root – corresponding to the names and 3D coordinates of edges and the connectivity amongst these edges. 



3. The animation path information can be provided by inputting the list of names of edges separated by space in the corresponding section. 



4. The histology images can be uploaded by providing images in PNG format and specifying the coordinates of images in the image upload section. 



5. The csv file which contains another set of coordinates of analysis can be provided in the last section if the user want to animate the transformation from 2D to 3D. 



Users can choose uploading a single cell csv file or a combination of several files and information to generate multiple kinds of visualization from a single-cell RNA-seq data to a narrative of spatial transcriptomic data.



To see an example of input data, please see the sampleData folder which contains example datasets based on previously published single-cell RNA-seq data, flow cytometry data and spatial transcriptomic data.



To easily visualize your data from other widely used single-cell analysis platforms, we provide the example scripts that can transform the Scanpy, Seurat, PAGA, Monocle3, Scaranoma results to the required input for StarmapVis.



## Scanpy



```sh

def adataTostarmap(adata, cord, featureN, valuetoplot, result_path):

    if cord == 'umap':

        cord_xyz = pd.DataFrame(adata.obsm['X_umap'],columns=['x', 'y'])

        cord_xyz['z'] = 1

    if cord == 'pca':

        cord_xyz = pd.DataFrame(adata.obsm['X_pca'][:,0:3], columns = ['x', 'y', 'z'])

    else:

        cord_x = pd.DataFrame(adata.obsm['X_pca'][:,int(cord.split("_")[1][0])-1], columns = ['x'])

        cord_y = pd.DataFrame(adata.obsm['X_pca'][:,int(cord.split("_")[1][1])-1], columns = ['y'])

        cord_z = pd.DataFrame(adata.obsm['X_pca'][:,int(cord.split("_")[1][2])-1], columns = ['z'])

        cord_xyz = pd.concat([cord_x,cord_y], axis=1, join='inner')

        cord_xyz = pd.concat([cord_xyz,cord_z], axis=1, join='inner')

        

    fcolumns = []

    features = {}

    for i in range(len(featureN)):

        fcolumns.append('PC'+str(featureN[i]))

        features[i] = list(adata.obsm['X_pca'][:,featureN[i]])

    pcs = pd.DataFrame.from_dict(features)

    pcs.columns = fcolumns

    if valuetoplot in adata.obs.columns:

        label = pd.DataFrame(adata.obs[valuetoplot])

        label['cluster'] = label[valuetoplot]

        label = label.reset_index()

    else:

        label = pd.DataFrame(adata.raw.obs_vector(valuetoplot).round(),columns = ['cluster'])

    adataStarmap = pd.concat([cord_xyz,pcs], axis=1, join='inner')

    adataStarmap = pd.concat([adataStarmap,label['cluster']], axis=1, join='inner')

    adataStarmap.to_csv(result_path,index = None)

```

Following is a example of using the function:

```sh

#Read the data from your scanpy result folder

adata = sc.read("~/write/pbmc3k.h5ad")

##Specify the corrdinates, features, cluster label and result path

    #cord can be 'umap', 'pca' or 'pca_'+numbers such as "pca_135", "umap" uses the umap values as 2D corrdinates, 'pca' takes pca_123 as 3D corrdinates, specific PC components are also acceptable.

cord = 'pca_135'

    #featureN takes a list of the number of PC components you want to use as features, max length is 12.

featureN = [1,2,3,4,5,6,7,8,9,10,11,12]

    #valuetoplot can be the same thing for Scanpy visualization, such as 'louvain' or Gene names

valuetoplot = 'louvain'

import scanpy as sc

import pandas as pd

result_path = '~/write/pbmc3k_starmap.csv'

adataTostarmap(adata, cord, featureN, valuetoplot, result_path)

```

An example can be found in /Convert from Scanpy or Seurat/Convert from Scanpy.ipynb .

## Seurat

```sh

library(dplyr)

library(Seurat)

#Read your data from Seurat result

pbmc <- readRDS(file = "filtered_gene_bc_matrices/output/pbmc_tutorial.rds")

pca_data <- pbmc[['pca']]@cell.embeddings

pca_data <- data.frame(pca_data)

#Select PC1,2,3 as 3D coordinates

pca_3d <- pca_data %>% select(1,2,3)

names(pca_3d)[1] <- "x"

names(pca_3d)[2] <- "y"

names(pca_3d)[3] <- "z"

#Select PC1,2,3,4,5,6,7,8,9,10,11,12 as 12 features

feature_data <- pca_data %>% select(1,2,3,4,5,6,7,8,9,10,11,12)

#Obtain the clustering result

cluster <- Idents(pbmc)

cluster <- data.frame(cluster)

pbmc_starmap <- bind_cols(pca_3d, feature_data)

pbmc_starmap <- bind_cols(pbmc_starmap, cluster)

#write the file into your result path

write.csv(pbmc_starmap,"filtered_gene_bc_matrices/output/pbmc_starmap.csv",row.names=F)

```

An example can be found in /Convert from Scanpy or Seurat/Convert from Seurat.ipynb .

## Spatial transcriptomics data

StarmapVR also accecpts spatial transcriptomics data processed by Scanpy.

```sh

def adataTostarmap_spatial(adata, featureN, valuetoplot, result_path):

    cord_xyz = adata.uns['spatial']['V1_Human_Lymph_Node']['scalefactors']['tissue_hires_scalef']*\

                pd.DataFrame(adata.obsm['spatial'],columns=['x', 'y'])

    cord_xyz['z'] = 1

    fcolumns = []

    features = {}

    for i in range(len(featureN)):

        fcolumns.append('PC'+str(featureN[i]))

        features[i] = list(adata.obsm['X_pca'][:,featureN[i]])

    pcs = pd.DataFrame.from_dict(features)

    pcs.columns = fcolumns

    

    if valuetoplot == 'leiden' or 'louvain':

        label = pd.DataFrame(adata.obs[valuetoplot])

        label['cluster'] = label[valuetoplot]

        label = label.reset_index()

    else:

        label = pd.DataFrame(adata.obs_vector(valuetoplot).round(),columns = ['cluster'])

    adataStarmap = pd.concat([cord_xyz,pcs], axis=1, join='inner')

    adataStarmap = pd.concat([adataStarmap,label['cluster']], axis=1, join='inner')

    adataStarmap.to_csv(result_path,index = None)

```

Zip the csv file with the tissue image, StarmapVR can visualize the spatial trancriptomic data in spatial dimensions. An example can be found in /Convert from Scanpy or Seurat/Convert from Scanpy (spatial).ipynb .The csv file and zip file should add ".sql" before the file extention. The example data can be found in /sampleData/humanlymphnode_4096_data.sql.zip.

## High-throghput imaging cytometry data

For image cytometry data with actual cell image, user can zip their cell images(cell index as the image name) with the requiring csv file, the csv file and zip file should add ".qpi" before the file extention. An example file can be found in /sampleData/Multi-ATOM_105kimage_data.qpi.zip.



## Usage instructions



starmapVis supports a number of input methods for interacting with the visualisation - keyboard, remote control and voice control. Note that voice control is available only in Google Chrome (desktop and mobile) as voice control utilises the SpeechRecognition API which is currently only supported by Chrome.



A summary of the control scheme for keyboard and voice control is included in the table below.



| Command | keyboard control | voice command |

| ------- | ---------------- | ------------- |

| forward | w | forward |

| backward | s | backward |

| left | a | left |

| right | d | right |

| zoom in | q | in |

| zoom out | e | out | 

| rotate Y-axis clockwise | left arrow | N\A |

| rotate Y-axis anti-clockwise | right arrow | rotate |

| rotate X-axis clockwise | up arrow | N\A |

| rotate X-axis anti-clockwise | down arrow | N\A |

| click on toolbox (VR mode) | N\A | select |

| reset toolbox (VR mode) | N\A | reset |

| reset view (VR mode) | N\A | init |