Closed howardchen0810 closed 3 years ago
Hi @howardchen0810 looks like the cells in your meta (celltype_tumor.txt
) don't match your columns in your counts (raw_data.txt
). Non of your count columns ends with _1
, _2
or _3
like your meta Cells. Could you double check that?
Hi, Thank you for your reply. From my count data file, I don't think there are cells ending with _1, _2 or _3. I just checked. Could you double check for me? I appreciate your help.
That's the issue! The cells on your meta don't match the columns on count, and they are supposed to.
oh! Sorry. I attached the wrong file. celltype.txt
This is the one I used. Could you run this one and check with me?
Here is the code and results I have.
Thanks!
@howardchen0810 You keep changing files and commands. It's hard to follow. Before I start to look at this again, can you confirm which version of CellPhoneDB and pandas are you using? pip show cellphonedb pandas
Here is the version I used.
I took both files you provided (celltype.txt
and raw_data.txt
) and ran cellphonedb with this command:
cellphonedb method statistical_analysis celltype.txt raw_data.txt --counts-data=gene_name
It took 20min and 18GB RAM but it didn't failed. Are you sure you're using the same dataset you shared?
I am 100% sure about using the same files I provided to you. It's probably because it ran too much RAM.
I'm not sure. That error probably comes from this chunk of code. But the files you shared shouldn't have a problem with those validations.
I could ask for your numpy
version, in case that counts.astype(np.float)
is failing, but that's a long shot. For example, you could try running this and see if you get an error:
import pandas as pd
import numpy as np
counts = pd.read_csv("raw_data.txt", sep="\t")
counts = counts.astype(np.float)
Hi,
I just finished the run with requesting more memory on server and it works!!!
I have another question when I run the dot_plot.
There is always have the error pop out about
Registered S3 methods overwritten by 'ggplot2':
method from
[.quosures rlang
c.quosures rlang
print.quosures rlang
Do you know how to fix it? Thanks
@howardchen0810 That's caused by having multiple packages with the same named functions. Should be more of a warning than an actual error. The last loaded package (ggplot2) is taking precedence for the function name and it's probably what you want when plotting.
Does that message prevents you from getting the plot output (i.e.: plot.pdf
)?
Plotting Registered S3 methods overwritten
warning is not related to "invalid counts" issue.
@howardchen0810 if that prevents you from getting your plot, please open a new ticket.
Hi Here is my count data. Becasue it is a big file, here is the link: https://drive.google.com/file/d/12fr_OYZSX4i-yLLcyc0C8UmT0vAoDKSl/view?usp=sharing here is my meta data. celltype_tumor.txt
Here is the code I ran: cellphonedb method analysis celltype.txt raw_data.txt --counts-data=gene_name
Could you help me to see what's wrong with my files? Thanks Howard