search

3DGenomes / TADbit

TADbit is a complete Python library to deal with all steps to analyze, model and explore 3C-based data. With TADbit the user can map FASTQ files to obtain raw interaction binned matrices (Hi-C like matrices), normalize and correct interaction matrices, identify and compare the so-called Topologically Associating Domains (TADs), build 3D models from the interaction matrices, and finally, extract structural properties from the models. TADbit is complemented by TADkit for visualizing 3D models
GNU General Public License v3.0
100 stars 61 forks source link

Problem while modelling #370

Open AngieD9110 opened 2 years ago

AngieD9110 commented 2 years ago

Hi, I am trying to test modelling with following command: tadbit model --workdir model_Ctrl_test --input_matrix test_Ctrl_matrix.f.bed --noX --optimize --beg 0 --end 20000000 --reso 50000 --maxdist 2000:4000:100 --nmodels 20 --nkeep 20 --exp_type Hi-C --cpu 32 --force However, an error occurred with the minimum maxdist value: image No matter how I adjust the minimum maxdist , the error keeps showing with only the minimum maxdist:(here I changed to 1000:4000:100) image Could you tell me how this happen? Many thanks!

david-castillo commented 2 years ago

Hi,

Are you using the last version of TADbit (just type tadbit version)? I remember a change back in May to keep compatibility with IMP where get_particle() was deprecated and that was changed to get().

Regards

David

AngieD9110 commented 2 years ago

Thanks so much for your response. It seems conda can only update to version 1.0. I updated the package using GitHub and it works. However, as I try to do downstream analysis(e.g. correlation real/models) with the following command: tadbit model --workdir model_Ctrl_test --input_matrix test_Ctrl_matrix.f.bed --noX --optimize --beg 0 --end 20000000 --reso 50000 --maxdist 1000:1100:100 --nmodels 20 --nkeep 20 --exp_type Hi-C --force --model --analyze --analyze_list 2 another problem occured: image

david-castillo commented 2 years ago

Hi,

Yes, sorry, that's a bit unclear in the documentation but you should run three different commands: one for the optimization where you infer the optimal parameters, one for the creation of the final ensemble of models and one to create the analysis plots. You can do several optimizations and tadbit will just add the combinations of parameters that haven't been tried in the previous runs. The idea is to maximize the correlation by trying different combinations of maxdist, lowfreq, upfreq and dcutoff. When you find the best correlation issue the tadbit model with --model.

tadbit model --optimize --workdir model_Ctrl_test --input_matrix test_Ctrl_matrix.f.bed --noX --beg 0 --end 20000000 --reso 50000 --maxdist 1000:1100:100 --nmodels 20 --nkeep 20 --exp_type Hi-C

tadbit model --model --workdir model_Ctrl_test --input_matrix test_Ctrl_matrix.f.bed --noX --beg 0 --end 20000000 --reso 50000 --nmodels 200 --nkeep 200 --exp_type Hi-C

tadbit model --analyze --workdir model_Ctrl_test

Regards

David