dtype issue with -R parameter

[hardingnj]

@hardingnj Hi, Nick, Seems I got similar dtype problem with the parameter -R ( --rrate ). My input variation data is vcf.gz format. My code is :

xpclr --format vcf -Sa "....." -Sb "....." -R 2.5e-06 -C 1 -I input.vcf.gz ... （this code worked for me in the old version）

It stopped after reading vcf and dropping some SNPs. It showed that : Traceback (most recent call last): File "/opt/miniconda2/bin/xpclr", line 195, in main() File "/opt/miniconda2/bin/xpclr", line 185, in main rrate=args.rrate) File "/opt/miniconda2/lib/python2.7/site-packages/xpclr/methods.py", line 287, in xpclr_scan geneticd = bpositions rrate File "/opt/miniconda2/lib/python2.7/site-packages/allel/abc.py", line 136, in mul return self.values other TypeError: ufunc 'multiply' did not contain a loop with signature matching types dtype('S11') dtype('S11') dtype('S11')

I changed 2.5e-06 to 1e-8 which is the same as the default. It still stopped there. I also changed it to 0.0000025, It still stopped there. When I delete the parameter -R and the rrate, it worked fine with the default rrate value 1e-08:

2019-11-29 16:43:24 : INFO : Done dropping above SNPs from analysis. XP-CLR algorithm starting. 2019-11-29 16:43:24 : INFO : No genetic distance provided; using rrate of 1e-08/bp 2019-11-29 16:43:25 : INFO : Omega estimated as : 0.355998

I tried to update numpy scipy pandas h5py and zarr to the newest version, but it didn't help.

Originally posted by @Dkyuan in https://github.com/hardingnj/xpclr/issues/43#issuecomment-559806844

[hardingnj]

Thanks, this certainly looks like a bug. Though not one I can immediately identify. Let me look into it!

[lcslyh]

Is this problem solved? I also report the same error.

xpclr -I ./PIGs.SNPs.chr.vcf -O ./Chr01.xpclr-LM.out -C chr1 -Sa ./pop.DL.table -Sb ./pop.DM.table --rrate 1e-8 --ld 0.95 --size 100000 --step 10000

2021-09-12 10:36:05 : INFO : Done dropping above SNPs from analysis. XP-CLR algorithm starting.
Traceback (most recent call last):
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/bin/xpclr", line 195, in <module>
    main()
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/bin/xpclr", line 181, in main
    model_lik, null_lik, selcoef, nsnps, navail, snpedges = xpclr.methods.xpclr_scan(
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/lib/python3.8/site-packages/xpclr/methods.py", line 283, in xpclr_scan
    geneticd = bpositions * rrate
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/lib/python3.8/site-packages/allel/abc.py", line 133, in __mul__
    return self.values * other
numpy.core._exceptions.UFuncTypeError: ufunc 'multiply' did not contain a loop with signature matching types (dtype('<U11'), dtype('<U11')) -> dtype('<U11')

[lcslyh]

When I delete the parameter -R and the rrate,It has a new error.

2021-09-12 12:36:55 : INFO : Omega estimated as : 0.062348
Traceback (most recent call last):
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/bin/xpclr", line 195, in <module>
    main()
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/bin/xpclr", line 181, in main
    model_lik, null_lik, selcoef, nsnps, navail, snpedges = xpclr.methods.xpclr_scan(
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/lib/python3.8/site-packages/xpclr/methods.py", line 320, in xpclr_scan
    weights = determine_weights(gt2.take(ix, axis=0), ldcutoff=ldcutoff,
  File "/home/lcs/Data/opt/biosoft/miniconda3/envs/xpclr/lib/python3.8/site-packages/xpclr/methods.py", line 270, in determine_weights
    above_cut = (squareform(ld**2) > ldcutoff) | (squareform(np.isnan(ld)))
TypeError: '>' not supported between instances of 'numpy.ndarray' and 'str'

[lcslyh]

After I delete the parameter --ld, it worked.

[hardingnj]

Thanks. Looks like the same bug:

https://github.com/hardingnj/xpclr/blob/a555c442b6ce9deff5ecff6e3080a5bde0acb557/bin/xpclr#L57

Seems that LD should be interpreted as a float, but something is not correct.