search

ChenWeiyan / LandSCENT

Landscape Single Cell Entropy
19 stars 6 forks source link

DoIntegPPI gene identifier error #2

Closed haircell closed 5 years ago

haircell commented 5 years ago

Hi,

I was able to run through the demo smoothly. I'm now trying a subset of my own data (SmartSeq2, about 900 cells). I'm using the SingleCellExperiment class with gene symbols. I was able to obtain the EntrezIDs through mapIds feature of AnnotationDbi. So I ultimately had EntrezIDs as row names (features/genes), and cell names as columns in my sce, which also contained additional metadata.

I'm looking to use the PPI network provided within the package, so both are loaded: data(net13Jun12.m) data(net17Jan16.m)

My data is in sce, when I try running the following: Integration.l <- DoIntegPPI(exp.m = sce, ppiA.m = net17Jan16.m, log_trans = FALSE) I get:

Error in DoIntegPPI(exp.m = sce, ppiA.m = net17Jan16.m, log_trans = FALSE) : 
  scRNA-seq data should have the same gene identifier with the network!

I get the same error with net13Jun12.m and when I try to use a matrix of either counts, normcounts, or logcounts all extracted from sce.

Here are the first 5x5 of my data:

                A1               A3              A4                   A5              A6
14679     0.00               0              0                   0.00           57.15
12544    55.97               0              0                   0.00            0.00
14955     0.00               0              0                   0.00            0.00
107815    0.00               0              0                   0.00            0.00
67608     0.00               0              0                   3.23           46.02

Any help would be appreciated. Thank you.

ChenWeiyan commented 5 years ago

Hi,

I checked the gene ID you provided, and they are all from mouse. However the networks we provided are all homo-sapiens genes, so I think the answer is clear.

I suggest two ways of solving this:

  1. You can find those homologous genes, could also be done with AnnotationDbi I think, and then use the network information we provided.
  2. You can also construct a network yourself with mouse genes and then do all the same workflow.

Sorry we do not highlight the species of network, I will add such information soon.

Hope this would be helpful.

Weiyan

haircell commented 5 years ago

Ah, that makes sense! Thanks.

Is there a workflow (or example) of how to create one of the PPI networks? I know the vignette mentions the sources it comes from, but would be great to have a work flow for updating purposes? Or a built in mouse network?

Thanks again!

ChenWeiyan commented 5 years ago

Yeah, we plan to add some examples of the network. May also consider providing other networks.

You could check relevant updates in the following versions.

cheers, Weiyan