csiu / promi2 Goto Github PK

Argument "" isn't numeric in multiplication (*) at promi2/code/external/tc_quantify/bin/rle_tpm_transform.pl line 48, <MFILE> line 184828.
Argument "" isn't numeric in multiplication (*) at promi2/code/external/tc_quantify/bin/rle_tpm_transform.pl line 48, <MFILE> line 184829.
...

Traceback (most recent call last):
  File "promi2/code/custom.py", line 436, in <module>
    main(args.f_config, args.infile, args.outdir, args.has_mirna, args.make_plots)
  File "promi2/code/custom.py", line 370, in main
    gff_allfeatures = extractFeatures_given_gff(f_config, gff_infile, outdir, has_mirna, is_consider_corr)
  File "promi2/code/custom.py", line 249, in extractFeatures_given_gff
    _, chrom, start, _, stop, strand = re.split('[r:.,]', l[3])
ValueError: need more than 1 value to unpack

Cause:

This is strange because there is fewer lines in the tc-infile (184827) then
what is referred to by the error (e.g. 184828+)
these extra lines become 0 in output: *max_tpm.bed

chr5  107718031  107718064  chr5:107718031..107718064,-  .  -  0
0     0          0          0                            0  0
0     0          0          0                            0  0

there are 6 columns of 0, program expects 7 columns

Input to custom.py should be unique w.r.t tss (or tss-mirna) position

Add verify feature?

custom.py: use max counts (current setup) or try all counts per positoin?

To find the probability of promoter per TSS position, we can

Use the max count of all samples to find prob(prom) ... O(1) per position
For each sample, find the prob(prom) and then take the max prob(prom) ... O(n) per position

Current setup is situation 1.

make sure directory does not exist for create-training-set.py

Traceback (most recent call last):
  File "create-training-set.py", line 664, in <module>
    main(args.files, args.outdir, args.n , args.l, args.get_id, args.f_config, args.verbose)
  File "create-training-set.py", line 533, in main
    os.path.join(outdir, 'tc-norm_negSet'))
  File "/raid2/local/python2.7.3/lib/python2.7/shutil.py", line 289, in move
    raise Error, "Destination path '%s' already exists" % real_dst
shutil.Error: Destination path '../Test-tset/tc-norm_negSet/tc-norm_negSet' already exists

make it so that if TSS is not within 50kb of miRNA, prediction will be background

Described label.py

add example?

Fix training.py so creation of training set does not depend on sets of infiles ...?

"ValueError: sample larger than population" when creating the training set

When creating the training set, we seed the background to N*10 positions (default N = 1000).
We then proceed to filter out positions which do not fit our requirements:

no repeats
no sequences near exon1
tc normalization (see tc_normalization.py)/overlap with peak
cpg <= 0.5 and cons <= 0.2 and tata <= 0.1
mirna_prox = 0

Per chance, the sample of positions left after filtering may be less than N ... causing the following error:

Traceback (most recent call last):
  File "create-training-set.py", line 660, in <module>
    main(args.files, args.outdir, args.n , args.l, args.get_id, args.f_config, args.verbose)
  File "create-training-set.py", line 578, in main
    selectedlines = random.sample(range(1,wc+1), N)
  File "/raid2/local/python2.7.3/lib/python2.7/random.py", line 320, in sample
    raise ValueError("sample larger than population")
ValueError: sample larger than population

Quick and easy thing to do is to rerun the code. Also make sure the tcnorm.ini is updated.

BUG: gff_unify is not unify based on miRNA position for mprox & corr

Getting way more positions than expected

csiu / promi2 Goto Github PK

promi2's People

Contributors

Watchers

Forkers

promi2's Issues

Recommend Projects

Recommend Topics

Recommend Org