Issue of deconvolution

[leihouyeung]

Thanks for your great work. I got an issue during deconvolution.

Error in enrich_matrix[, ct_spot_k[m]] : subscript out of bounds
In addition: Warning messages:
1: In sparseMatrix(i = indices[] + 1, p = indptr[], x = as.numeric(x = counts[]), :
'giveCsparse' has been deprecated; setting 'repr = "T"' for you
2: In cbind(allCounts_DWLS, solDWLS) :
number of rows of result is not a multiple of vector length (arg 2)

And I also notice that there is no guideline for deconvolution in the tutorial of Giotto. Could you supply this for me?

[JunxiangXu]

Sorry for the late reply. I am unable to replicate this error message you provided. Can you specify which command did you get the error message from? Can you also upgrade to the latest Giotto Suite version and see if it still appears?

Also, We have a tutorial on deconvolution/spatial enrichment at http://giottosuite.com/articles/analyses_deconvolution_Oct2021.html, or you can refer to the complete pipeline of visium brain dataset: http://giottosuite.com/articles/mouse_visium_brain_220426.html.

[leihouyeung]

Thanks for your reply. It still didn't work on my dataset by outputting this error when I run "runDWLSDeconv":

Error in quadprog::solve.QP(Dmat = D, dvec = d, Amat = A, bvec = bzero) :
matrix D in quadratic function is not positive definite!

[gcyuan]

This seems to be a data rather than software issue. Have you tried other datasets to see if the software works?

…

On Apr 30, 2022, at 9:56 AM, leihouyeung ***@***.***> wrote:  Thanks for your reply. It still didn't work on my dataset by outputting this error when I run "runDWLSDeconv": Error in quadprog::solve.QP(Dmat = D, dvec = d, Amat = A, bvec = bzero) : matrix D in quadratic function is not positive definite! — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.

[leihouyeung]

There is also another error when I applied the other dataset:

Error in eigen(if (doDykstra) R else Y, symmetric = TRUE) :
infinite or missing values in 'x'

These two datasets could be deconvolved successfully by some other methods. So I'm really confused.

On the other side, could you supply a more concise tutorial for generalized data format (like I have :1. scRNA-seq expression matrix, 2. cell-type annotation of all cells, 3. ST expression matrix and 4. coordinates of all ST spots). Because maybe your tutorial would work for your dataset, but if I want to apply this method on my dataset, it would be hard to transfer your data format to mine. That's just a suggestion. :)

[gcyuan]

Have you tried following the steps described in our Current Protocols paper? Here is the link for the R markdown: https://github.com/drieslab/giotto_current_protocols And the paper; https://currentprotocols.onlinelibrary.wiley.com/doi/10.1002/cpz1.405

…

On Apr 30, 2022, at 1:02 PM, leihouyeung ***@***.***> wrote:  There is also another error when I applied the other dataset: Error in eigen(if (doDykstra) R else Y, symmetric = TRUE) : infinite or missing values in 'x' These two datasets could be deconvolved successfully by some other methods. So I'm really confused. On the other side, could you supply a more concise tutorial for generalized data format (like I have :1. scRNA-seq expression matrix, 2. cell-type annotation of all cells, 3. ST expression matrix and 4. coordinates of all ST spots). Because maybe your tutorial would work for your dataset, but if I want to apply this method on my dataset, it would be hard to transfer your data format to mine. That's just a suggestion. :) — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.

[leihouyeung]

@gcyuan Thanks for your supply, but I still have the mentioned error when running deconvolution. I have tried too many versions of Giotto for deconvolution. How about I send you the data and could you please help me apply your method for deconvolution? The link is here.
object$sc_count expression profile of scRNA-seq
object$st_count expression profile of ST
object$cell_type cell type annotation of scRNA-seq
object$st_location coordinates of ST spots

Thanks so much!

[gcyuan]

Sure. Happy to help.

…

On Apr 30, 2022, at 2:08 PM, leihouyeung ***@***.***> wrote:  @gcyuan Thanks for your supply, but I still have the mentioned error when running deconvolution. I have tried too many versions of Giotto for deconvolution. How about I send you the data and could you please help me apply your method for deconvolution? The link is here. object$sc_count expression profile of scRNA-seq object$st_count expression profile of ST object$cell_type cell type annotation of scRNA-seq object$st_location coordinates of ST spots Thanks so much! — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.

[gcyuan]

Hi This tip from our former Member Natalie del Rossi may help. ‘I don’t know if the error is related to anything specific, as it seems like it could be caused by a number of things. My guess is that it might be an OS complication on the user’s side. This could be a good resource for them to look into:rocker-org/rocker-versioned2#259.’ Guo-Cheng

…

On Apr 30, 2022, at 2:08 PM, leihouyeung ***@***.***> wrote:  @gcyuan Thanks for your supply, but I still have the mentioned error when running deconvolution. I have tried too many versions of Giotto for deconvolution. How about I send you the data and could you please help me apply your method for deconvolution? The link is here. object$sc_count expression profile of scRNA-seq object$st_count expression profile of ST object$cell_type cell type annotation of scRNA-seq object$st_location coordinates of ST spots Thanks so much! — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.

[leihouyeung]

@gcyuan Have you tried our data on SpatialDWLS? If you deconvolved it successfully, you could share your code with me so that I could see why my code did not work. Thanks. From the report of the error, I do not really think it would be related to the OS complication.

[curiegat]

Hi @leihouyeung and @gcyuan

May I know whether the problems related to the Eigen error (Error in eigen(if (doDykstra) R else Y, symmetric = TRUE) : infinite or missing values in 'x' - #277) has been solved or not?, if it has, may I know the solution?, because currently I am working with 5 samples and 5 gene lists, and I got this error message on 4 samples with 2 gene lists. I have tried to omit and remove the NAs, but it seems like not giving any impact. I would be glad if someone could help and I'll share the data that have the error.

Thank you and cheers,

Agatha

[chrkuo]

There is also another error when I applied the other dataset:

Error in eigen(if (doDykstra) R else Y, symmetric = TRUE) : infinite or missing values in 'x'

These two datasets could be deconvolved successfully by some other methods. So I'm really confused.

On the other side, could you supply a more concise tutorial for generalized data format (like I have :1. scRNA-seq expression matrix, 2. cell-type annotation of all cells, 3. ST expression matrix and 4. coordinates of all ST spots). Because maybe your tutorial would work for your dataset, but if I want to apply this method on my dataset, it would be hard to transfer your data format to mine. That's just a suggestion. :)

I am running DWLS and just encountered the same error. do you mind sharing what other methods helped with deconvolution to bypass this error? and what was the cause of that error?