BINDetect ValueError

Hi @msbentsen, I have tried using TOBIAS v0.12.1 BINDetect across 25 samples with this code,

TOBIAS BINDetect --signals example/*_corrected_footprints.bw --motifs example/motifs --genome example/fasta --peaks example/bed --skip-excel --outdir output/bindetect --verbosity 4 --cores 25

and I keep running into this error

/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/tobias/tools/bindetect.py:388: UserWarning: Creating legend with loc="best" can be slow with large amounts of data.
  figure_pdf.savefig(fig, bbox_inches='tight')
Traceback (most recent call last):
  File "/home/jopham/anaconda3/envs/TOBIAS/bin/TOBIAS", line 8, in <module>
    sys.exit(main())
  File "/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/tobias/TOBIAS.py", line 154, in main
    args.func(args)
  File "/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/tobias/tools/bindetect.py", line 412, in r
un_bindetect
    gmm.fit(np.log(bg_values).reshape(-1, 1))
  File "/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/sklearn/mixture/_base.py", line 193, in fit
    self.fit_predict(X, y)
  File "/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/sklearn/mixture/_base.py", line 220, in fit_predict
    X = _check_X(X, self.n_components, ensure_min_samples=2)
  File "/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/sklearn/mixture/_base.py", line 53, in _check_X
    ensure_min_samples=ensure_min_samples)
  File "/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/sklearn/utils/validation.py", line 72, in inner_f
return f(**kwargs)
  File "/home/jopham/anaconda3/envs/TOBIAS/lib/python3.7/site-packages/sklearn/utils/validation.py", line 653, in check_array
    context))
ValueError: Found array with 0 sample(s) (shape=(0, 1)) while a minimum of 2 is required.

I have checked that the *_corrected_footprints.bw files have information in them. Also, I have the log file if you would like it as well.

Sorry about the formatting of the error, this is my first time commenting an issue on github and thanks for helping in advance!

[msbentsen]

Thank you for all of the information @parmejohn and @blairperry 😃 I narrowed it down to the across-condition normalization within BINDetect, which failed for some specific outlier cases, thereby introducing NA’s into the data. I made some changes in tobias 0.12.4 to make this more robust. The normalization is a little different, so for the sake of comparison, please update and rerun analysis (even those which might have run through with the old version).

I will close this issue, but thank you for your inputs, and please reopen if you encounter any other issues!

[blairperry]

Despite the error when running the example data, BINDetect worked perfectly when I plugged in my own data and ran commands manually instead of within the snakemake pipeline. I’m not sure if this is helpful, but figured it might help narrow in on the issue.