comparison RepEnrich_setup.py @ 15:2e3d976e7d5d draft default tip

planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/main/tools/repenrich commit 03183e29f807ec33548016a7c4144f52720b7b9e

14:bf866bedd4b4

# load genome into dictionary and compute length
g = SeqIO.to_dict(SeqIO.parse(genomefasta, "fasta"))
genome = defaultdict(dict)

for chr in g.keys():
    genome[chr]['sequence'] = g[chr].seq
    genome[chr]['length'] = len(g[chr].seq)

# Build a bedfile of repeatcoordinates to use by RepEnrich region_sorter
repeat_elements = set()
rep_coords = defaultdict(list)  # Merged dictionary for coordinates

# generate spacer for pseudogenomes
spacer = ''.join(['N' for i in range(gapl)])

# generate metagenomes and save them to FASTA files for bowtie build
for repname in rep_coords:
    metagenome = ''
    # iterating coordinate list by block of 3 (chr, start, end)
    block = 3
    for i in range(0, len(rep_coords[repname]) - block + 1, block):
        batch = rep_coords[repname][i:i+block]
        chromosome = batch[0]
        start = max(int(batch[1]) - flankingl, 0)
        end = min(int(batch[2]) + flankingl,
                  int(genome[chromosome]['length'])-1) + 1
        metagenome = (
            f"{metagenome}{spacer}"
            f"{genome[chromosome]['sequence'][start:end]}"
            )

    # Create Fasta of repeat pseudogenome
    fastafilename = f"{repname}.fa"
    record = SeqRecord(Seq(metagenome), id=repname, name='', description='')
    SeqIO.write(record, fastafilename, "fasta")

15:2e3d976e7d5d

# load genome into dictionary and compute length
g = SeqIO.to_dict(SeqIO.parse(genomefasta, "fasta"))
genome = defaultdict(dict)

for chr in g.keys():
    genome[chr]['sequence'] = str(g[chr].seq)
    genome[chr]['length'] = len(g[chr].seq)

# Build a bedfile of repeatcoordinates to use by RepEnrich region_sorter
repeat_elements = set()
rep_coords = defaultdict(list)  # Merged dictionary for coordinates

# generate spacer for pseudogenomes
spacer = ''.join(['N' for i in range(gapl)])

# generate metagenomes and save them to FASTA files for bowtie build
for repname in rep_coords:
    genomes_list = []
    # iterating coordinate list by block of 3 (chr, start, end)
    block = 3
    for i in range(0, len(rep_coords[repname]) - block + 1, block):
        batch = rep_coords[repname][i:i+block]
        chromosome = batch[0]
        start = max(int(batch[1]) - flankingl, 0)
        end = min(int(batch[2]) + flankingl,
                  int(genome[chromosome]['length'])-1) + 1
        genomes_list.append(genome[chromosome]['sequence'][start:end])
    metagenome = spacer.join(genomes_list)
    # Create Fasta of repeat pseudogenome
    fastafilename = f"{repname}.fa"
    record = SeqRecord(Seq(metagenome), id=repname, name='', description='')
    SeqIO.write(record, fastafilename, "fasta")