estProp always even across types

[umasstr]

To test CDSeq, I used counts from published bulk RNA-seq of human skin (accession numbers below). Gene-level counts were used as input. Surprisingly, estProp from the output always distributes the proportion of single cells evenly across the types. the below command runs to completion. Is there a reason that this might happen? A similar result is also returned with subsets of these samples.

cdsrr<-CDSeq(bulk_data =  srr, 
             cell_type_number = 10, 
             mcmc_iterations = 5, # increase the mcmc_iterations to 700 or above
             gene_length = as.vector(length),
             cpu_number = 6)

SRR12330925 SRR12330926 SRR12330927 SRR12330928 SRR12330929 SRR12330930 SRR12330931 SRR12330932 SRR12330933 SRR12330947 SRR12330948 SRR12330949 SRR12330950 SRR12330951 SRR12330952 SRR12330953

[kkang7]

it seems the mcmc runs are only 5? if that's the case, you should run more mcmcs, say 700 or above.

On Sat, Apr 29, 2023 at 8:05 PM GitDog @.***> wrote:

To test CDSeq, I used counts from published bulk RNA-seq of human skin (accession numbers below). Gene-level counts were used as input. Surprisingly, estProp from the output always distributes the proportion of single cells evenly across the types. the below command runs to completion. Is there a reason that this might happen? A similar result is also returned with subsets of these samples.

cdsrr<-CDSeq(bulk_data = srr, cell_type_number = 10, mcmc_iterations = 5, # increase the mcmc_iterations to 700 or above gene_length = as.vector(length), cpu_number = 6)

SRR12330925 SRR12330926 SRR12330927 SRR12330928 SRR12330929 SRR12330930 SRR12330931 SRR12330932 SRR12330933 SRR12330947 SRR12330948 SRR12330949 SRR12330950 SRR12330951 SRR12330952 SRR12330953

— Reply to this email directly, view it on GitHub https://github.com/kkang7/CDSeq_R_Package/issues/22, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEDIUAN4FIFH25U4GSUN6K3XDWUGHANCNFSM6AAAAAAXQQ464M . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[umasstr]

Thanks for your reply. I repeated the above command, setting the mcmc_iterations argument = 700 and still get the same results. I also tried setting cell_type_number = 5:10, which produced 5 celltypes, all with the exact same proportions

[kkang7]

what is the total counts for each bulk? say what is colSums(srr)?