Commit 64f80f3b authored by Pavel Jedlicka's avatar Pavel Jedlicka
Browse files

LTRharvest implemented; can add more discovery tools

parent ec377525

.gitignore

+5 −0
Original line number Original line Diff line number Diff line
@@ -7,3 +7,8 @@ tmp 
images

images

results

results

config.yml

config.yml

LTR_Finder/

build/

nested.egg-info/

dist/

.vscode/

config_template.yml

+11 −1
Original line number Original line Diff line number Diff line
@@ -15,6 +15,16 @@ ltr: 
    a: *prosite_path

    a: *prosite_path

    s: *tRNAdb_path

    s: *tRNAdb_path





ltr_harvest:

  path: 'gt'

  command: 'ltrharvest'

  args:

    similar: 40

    maxlenltr: 3000

    maxdistltr: 20000

    mintsd: 4

    maxtsd: 10

       

gt:

gt:

  path: 'gt'

  path: 'gt'

  command: 'sketch'

  command: 'sketch'

nested/cli/nester.py

+18 −9
Original line number Original line Diff line number Diff line
@@ -12,7 +12,8 @@ from Bio import SeqIO 




from nested.core.nester import Nester

from nested.core.nester import Nester

from nested.output.sketcher import Sketcher

from nested.output.sketcher import Sketcher

from nested.core.sequence_thread import SequenceThread

from nested.core.discovery_tool import DiscoveryTool

from nested.utils.dependency_resolver import DependencyResolver









@click.command()

@click.command()
@@ -24,28 +25,36 @@ from nested.core.sequence_thread import SequenceThread 
@click.option('--initial_threshold', '-t', type=int, default=500, help='Initial threshold value.')

@click.option('--initial_threshold', '-t', type=int, default=500, help='Initial threshold value.')

@click.option('--threshold_multiplier', '-m', type=float, default=1, help='Threshold multiplier.')

@click.option('--threshold_multiplier', '-m', type=float, default=1, help='Threshold multiplier.')

@click.option('--threads', '-n', type=int, default=1, help='Number of threads')

@click.option('--threads', '-n', type=int, default=1, help='Number of threads')

def main(input_fasta, sketch, format, output_fasta_offset, output_folder, initial_threshold, threshold_multiplier, threads):    

@click.option('--discovery_tool', '-dt', default=DiscoveryTool.LTR_finder.name,

    type=click.Choice(list(map(lambda val: val, DiscoveryTool.__members__))),

    help='Determines which tool is used for retrotransoson discovery. Default: LTR_finder')

def main(input_fasta, sketch, format, output_fasta_offset, output_folder, initial_threshold, 

        threshold_multiplier, threads, discovery_tool):

    check_permissions(output_folder, os.W_OK | os.X_OK)

    check_permissions(output_folder, os.W_OK | os.X_OK)

    number_of_errors = [0]

    number_of_errors = [0]

    start_time = datetime.now()

    start_time = datetime.now()

    sequences = list(SeqIO.parse(open(input_fasta), 'fasta'))

    sequences = list(SeqIO.parse(open(input_fasta), 'fasta'))

    sketcher = Sketcher()  

    sketcher = Sketcher()  

    futuress = []

    futuress = []

    dependencyResolver = DependencyResolver(discovery_tool)

    with futures.ThreadPoolExecutor(threads) as executor:

    with futures.ThreadPoolExecutor(threads) as executor:

        for sequence in sequences:

        for sequence in sequences:

            futuress.append(executor.submit(process_sequence, sequence, sketcher, sketch, format, output_fasta_offset, output_folder, initial_threshold, threshold_multiplier, number_of_errors))

            futuress.append(executor.submit(process_sequence, sequence, sketcher, 

                sketch, format, output_fasta_offset, output_folder, 

                initial_threshold, threshold_multiplier, number_of_errors,

                dependencyResolver))

        futures.wait(futuress)

        futures.wait(futuress)

    end_time = datetime.now()

    end_time = datetime.now()

    print('Total time: {}'.format(end_time - start_time))

    print('Total time: {}'.format(end_time - start_time))

    print('Number of errors: {}'.format(number_of_errors[0]))

    print('Number of errors: {}'.format(number_of_errors[0]))





def process_sequence(sequence, sketcher, sketch, format, output_fasta_offset, output_folder, initial_threshold, threshold_multiplier, errors):

def process_sequence(sequence, sketcher, sketch, format, output_fasta_offset, output_folder, initial_threshold, threshold_multiplier, errors, dependency_resolver):

    sequence.id = sequence.id.replace('/', '--')

    sequence.id = sequence.id.replace('/', '--')

    seq_start_time = datetime.now()

    seq_start_time = datetime.now()

    strlen = 15

    strlen = 15

    print("Processing {}".format(sequence.id))

    print("Processing {}".format(sequence.id))

    try:

    try:

        nester = Nester(sequence, initial_threshold, threshold_multiplier)

        nester = Nester(sequence, initial_threshold, threshold_multiplier, dependency_resolver)

        sketcher.create_gff(nester.nested_element, dirpath=output_folder, output_fasta_offset=output_fasta_offset, format=format)

        sketcher.create_gff(nester.nested_element, dirpath=output_folder, output_fasta_offset=output_fasta_offset, format=format)

        sketcher.create_solo_ltr_gff(nester.solo_ltrs, dirpath=output_folder)  

        sketcher.create_solo_ltr_gff(nester.solo_ltrs, dirpath=output_folder)  

        if sketch:

        if sketch:
@@ -57,14 +66,14 @@ def process_sequence(sequence, sketcher, sketch, format, output_fasta_offset, ou 
    except KeyboardInterrupt:

    except KeyboardInterrupt:

        cleanup()

        cleanup()

        raise

        raise

    except CalledProcessError:

    except CalledProcessError as ex:

        cleanup()

        cleanup()

        errors[0] += 1

        errors[0] += 1

        print('Processing {}: SUBPROCESS ERROR'.format(sequence.id[:strlen]))

        print('Processing {}: SUBPROCESS ERROR: {}'.format(sequence.id[:strlen], ex))

    except Exception:

    except Exception as ex:

        cleanup()

        cleanup()

        errors[0] += 1

        errors[0] += 1

        print('Processing {}: UNEXPECTED ERROR:'.format(sequence.id[:strlen]), sys.exc_info()[0])

        print('Processing {}: UNEXPECTED ERROR: {}'.format(sequence.id[:strlen], ex))





def check_permissions(path, permissions):

def check_permissions(path, permissions):

    path = os.getcwd() if len(path) == 0 else path

    path = os.getcwd() if len(path) == 0 else path

nested/core/discovery_tool.py

0 → 100644
+11 −0
Original line number Original line Diff line number Diff line 
#!/usr/bin/env python3



from enum import Enum



class DiscoveryTool(Enum):

    LTR_finder = 1

    LTRharvest = 2

    #add aliases with same value as the tool above

    finder = 1

    harvest = 2

nested/core/gene.py

+5 −3
Original line number Original line Diff line number Diff line 
#!/usr/bin/env python3

#!/usr/bin/env python3





from nested.core.te import run_ltr_finder

#from nested.core.te import run_ltr_finder

# from nested.core.domain import run_blastx

# from nested.core.domain import run_blastx

# from nested.core.domain_dante import run_last

# from nested.core.domain_dante import run_last

from nested.core.domain_dante import run_blastx

from nested.core.domain_dante import run_blastx

from nested.scoring.graph import Graph

from nested.scoring.graph import Graph

from nested.utils import intervals

from nested.utils import intervals

from nested.utils.dependency_resolver import DependencyResolver









class Gene(object):

class Gene(object):
@@ -20,10 +21,11 @@ class Gene(object): 




    """

    """





    def __init__(self, seqid=None, sequence=None):

    def __init__(self, dependency_resolver, seqid=None, sequence=None):

        self.seqid = seqid

        self.seqid = seqid

        self.sequence = sequence

        self.sequence = sequence

        self.te_list = run_ltr_finder(seqid, sequence)

        self.te_list = dependency_resolver.discovery_tool.run(seqid, sequence)

        #self.te_list = run_ltr_finder(seqid, sequence)

        self.domain_list = run_blastx(sequence)

        self.domain_list = run_blastx(sequence)

        # self.domain_list = run_last(sequence)

        # self.domain_list = run_last(sequence)

        self.scores = []

        self.scores = []