search

brycerowland / thundeR

0 stars 0 forks source link

Why nothing returned from function thunder_feature_selection ? #1

Open junmeiW opened 3 years ago

junmeiW commented 3 years ago

Hi, I'm very interested in the application of this package, so I tried the thundeR package on my dataset.

For example， I had some mouse bulk Hi-C samples mixed with two cell types with known proportions, and I first used thunder_feature_selection the step one of THUNDER algorithm for each chromosome (Chromosome 1-Chromosome X). But I got nothing from this function for every chromosome data. It is a small part of my data for chromosome 10 as below(these are KR normalized values from juicer tools) :

bin_name bulk_7 bulk_8 bulk_9 bulk_70
3000000_3000000 18442.8770 28298.6170 25003.6230 24095.7130
3000000_3500000 4540.7544 7692.3220 6381.2007 6176.3433
3000000_4000000 2859.7800 4687.5293 4002.2837 4064.0886
3000000_4500000 1148.5991 1916.2637 1638.8638 1611.6409
3000000_5000000 804.9667 935.7129 926.6263 916.0406

And the code was like:

## chr10_dat = readr::read_tsv("../data/g1_es_chr10.tsv")
fea_sele <- thunder_feature_selection("../data/samples_chr10.tsv",
                                      # subset_mix_out_path = "../data/samples_sub_chr10.tsv",
                                      n_cell_types = 2)
fea_sele

and the fea_sele returns as below 0 rows | 1-10 of 12 columns

The problems happened with all chromosomes and even other bulk samples with other species. So I wonder why this problem happens and how to fix it.

I'm looking forward to your reply sincerely! Thank you!

brycerowland commented 2 years ago

My apologies for the delay in responding. My guess is that the input data was not correctly formatted. If you are still interested in using our package, please check the package documentation in the README.md, and please let me know if you have any additional questions.

junmeiW commented 2 years ago

Thanks for your reply. I've changed the input data with correct format, and it runs successfully. And I wander how to know the proportions returned from the get_props() corresponding to which cell type?

Results below are returned from function get_props() for some mixtures, there are 2 columns(2 cell types, for example A cell type and B cell type), how to know column prop1 or prop2 is the estimated proportion of A cell type ? Mix_sampleID prop1 prop2
mix_7 1.891126e-15 1.0000000000
mix_8 9.998815e-01 0.0001185301
mix_9 4.751779e-01 0.5248220718
mix_70 4.249946e-01 0.5750054313