Error when load data by readVisium

[hannshu]

Hi,

Thank you for your excellent work on BayesSpace. I am currently attempting to implement BayesSpace with a 10x Visium data link, I've downloaded the Feature / barcode matrix (filtered) file and the Spatial imaging data file, the file are organized like this

.
├── filtered_feature_bc_matrix
│   ├── barcodes.tsv.gz
│   ├── features.tsv.gz
│   └── matrix.mtx.gz
├── filtered_feature_bc_matrix.h5
├── Parent_Visium_Human_BreastCancer_filtered_feature_bc_matrix.tar.gz
└── spatial
    ├── aligned_fiducials.jpg
    ├── detected_tissue_image.jpg
    ├── scalefactors_json.json
    ├── tissue_hires_image.png
    ├── tissue_lowres_image.png
    └── tissue_positions_list.csv

Then I try to load the data by the readVisium function like this

library(BayesSpace)

hbc <- readVisium("hbc/")

It prompt error

Error in .subscript.2ary(x, , j, drop = TRUE) : subscript out of bounds
Calls: readVisium -> [ -> [ -> .subscript.2ary
Execution halted

I find a function read10Xh5 in your source code and try to use this function to load data. This function can successfully load the data but it prompts error in the spatialCluster function

Error: subscript contains invalid names
Execution halted

I'm a novice in R, my following script is like this, which is modified from your tutorial on the DLPFC dataset link

hbc <- read10Xh5("hbc/")
hbc <- logNormCounts(hbc)

set.seed(101)
dec <- scran::modelGeneVar(hbc)
top <- scran::getTopHVGs(dec, n = 2000)

set.seed(102)
hbc <- scater::runPCA(hbc, subset_row=top)

## Add BayesSpace metadata
hbc <- spatialPreprocess(hbc, platform="Visium", skip.PCA=TRUE)

q <- 16  # Number of clusters
d <- 20  # Number of PCs

## Run BayesSpace clustering
set.seed(104)
hbc <- spatialCluster(hbc, q=q, d=d, platform='Visium',
                        nrep=50000, gamma=3, save.chain=TRUE)

Please help me solve this problem.

[kunstner]

Same problem here. Any progress on this issue?

[g-wee]

Another novice with the same problem here. Hoping for a solution soon.

[edward130603]

Sorry for the late response. This issue should be fixed in the next bioconductor release. You can also use the current devel branch on github as well.