Error in res_imp$imputed[gene.expression == 0]

[ansonrel]

Hi,

I am trying VIPER on different datasets but it fails for one of them with the following error message:

Error in res_imp$imputed[gene.expression == 0] :
  invalid subscript type 'S4'

.. with the default parameters. I tried with an alpha of 0.5 and 1 to see if it is dependent on the method used but it leads to the same error. I also tried to feed in raw counts as well as CPM.

The dataset is of size 16k features x 900 cells, not that big, so I guess it shouldn't be an issue either. The datasets that worked have a similar size.

And as for #3, the function runs really slowly on my datasets (SMARTer and 10x protocol).

Do you have an idea to resolve this ?

Thanks,
Anthony

[ChenMengjie]

Hi Anthony, Can you check the zero rate of your datasets? It looks like after filtering, the matrix becomes empty. Maybe some rows or columns have all zeros. I know it is slow on 10X datasets. We now don’t have implementation that optimizes for low coverage large number datasets. Also our method fits regression model based on log transformed datasets. For some UMI, it might not be the optimal choice. Thanks, Mengjie On Sep 25, 2019, at 2:28 AM, ansonrel <notifications@github.com<mailto:notifications@github.com>> wrote: Hi, I am trying VIPER on different datasets but it fails for one of them with the following error message: Error in res_imp$imputed[gene.expression == 0] : invalid subscript type 'S4' .. with the default parameters. I tried with an alpha of 0.5 and 1 to see if it is dependent on the method used but it leads to the same error. I also tried to feed in raw counts as well as CPM. The dataset is of size 16k features x 900 cells, not that big, so I guess it shouldn't be an issue either. The datasets that worked have a similar size. And as for #3<#3>, the function runs really slowly on my datasets (SMARTer and 10x protocol). Do you have an idea to resolve this ? Thanks, Anthony — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#4?email_source=notifications&email_token=AD7DL3NB2HASV7TR5HZKD4LQLMHJNA5CNFSM4I2I22U2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HNQNRFA>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AD7DL3PK5UUQOZ7RA6FLOJ3QLMHJNANCNFSM4I2I22UQ>.

[ansonrel]

Thanks for your quick answer Mengjie.
Apparently the dataset contains no empty row/column, and there are approximately 46% of zeros in the matrix.

[ChenMengjie]

How soon did you get the warning? Did it run for a while before the warning?

…

Sent from my iPhone On Sep 26, 2019, at 12:54 AM, ansonrel <notifications@github.com<mailto:notifications@github.com>> wrote: Thanks for your quick answer Mengjie. Apparently the dataset contains no empty row/column, and there are approximately 46% of zeros in the matrix. — You are receiving this because you commented. Reply to this email directly, view it on GitHub<#4?email_source=notifications&email_token=AD7DL3IM3I7A4NPKGRAIDXLQLRE7FA5CNFSM4I2I22U2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD7UMMBI#issuecomment-535348741>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AD7DL3OPD4ATQODDDUKVDBTQLRE7FANCNFSM4I2I22UQ>.

[ansonrel]

Indeed, it ran for several (>2) hours before I got the message.