Generating a sample annotation file for inferCNV

[Alaine]

Hi there,

I am a new user of inferCNV, thank you for this tool!
I have several datasets that I want to run inferCNV on, but I don't have sample annotation files which specify cell type for them.
How should I go about preparing a sample cell type annotation file for these datasets, so that I may run inferCNV on them?

Thank you in advance!

[Christophe Georgescu]

Hi Alaine,

Do you already have information on what cell types you have and their status (healthy/malignant), meaning you simply need to generate a file that contains the information properly formatted, or do you need to do cell type assignments first?

Regards,

Christophe.

[Alaine]

Hi Christophe,

I need to do cell type assignments first, I was wondering what method would be best to do this.

Thanks,

Alaine

[Christophe Georgescu]

Hi Alaine,

You can take a look at this benchmark of different methods to decide based on the size of your dataset, its type and other experimental knowledge which one should be best fitting : https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1795-z . Another tool that could be worth taking a look (published on the same day so not part of the benchmark) at is https://www.nature.com/articles/s41592-019-0529-1 .

Regards,

Christophe.

[Alaine]

Hi Christophe,

 

Thank you for your previous response. I would like to clarify on what point. In our study, we are interested in using InferCNV to detect malignant cells. Identifying different cell types within the dataset is not the focus of our study. However, in order to use InferCNV, we need to generate an annotation file in an unbiased way. I understand that you have suggested that we use classification methods proposed in the above papers to generate an annotation file. We have tried those. However, there are two issues with using supervised methods:

1. In supervised methods, the cell types should be known a priori; while this is not the case for many of our datasets. That is, we do not have cell labels for many datasets.
2. Generating an independent annotation file by using cell labels from the same dataset does not seem to be the correct way to do this? Please correct me, if this is not true. 

Would you recommend any other way to generate annotation files?

Additionally, I had a question regarding the gene expression matrix input for InferCNV. It says that we should use a matrix of raw counts, however some of my datasets are in CPM or TPM. Can I apply inferCNV to datasets in TPM or CPM? 

 

Thank you,

Alaine

[vrutant shah]

Hi, Were you able to generate annotation file? 

if yes, can you please share how to do that?