Mercurial > repos > devteam > fastq_manipulation
diff fastq_manipulation.xml @ 7:3b9ecb0920bf draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tool_collections/galaxy_sequence_utils/fastq_manipulation commit a5766d27dcddd1891766476a913d0eae1ec7a3c9
| author | iuc |
|---|---|
| date | Sun, 23 Nov 2025 17:48:47 +0000 |
| parents | ed6c5df75175 |
| children |
line wrap: on
line diff
--- a/fastq_manipulation.xml Fri Oct 04 10:33:21 2024 +0000 +++ b/fastq_manipulation.xml Sun Nov 23 17:48:47 2025 +0000 @@ -20,13 +20,8 @@ #import binascii import binascii import re -import six -if six.PY2: - from string import maketrans -else: - maketrans = str.maketrans - +maketrans = str.maketrans ##does read match def match_read(fastq_read): @@ -253,162 +248,346 @@ <!-- match all and do nothing --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="identifier" /> - <param name="manipulation_selector" value="translate" /> - <param name="from" value="" /> - <param name="to" value="" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="identifier" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="translate" /> + <param name="from" value="" /> + <param name="to" value="" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> </test> <!-- match None and do nothing --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value="STRINGDOESNOTEXIST" /> - <param name="manipulation_type_selector" value="identifier" /> - <param name="manipulation_selector" value="translate" /> - <param name="from" value="" /> - <param name="to" value="" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value="STRINGDOESNOTEXIST" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="identifier" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="translate" /> + <param name="from" value="" /> + <param name="to" value="" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and remove --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="miscellaneous" /> - <param name="manipulation_selector" value="remove" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="miscellaneous" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="remove" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="empty_file.dat" ftype="fastqsanger" /> </test> <!-- match None and remove --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value="STRINGDOESNOTEXIST" /> - <param name="manipulation_type_selector" value="miscellaneous" /> - <param name="manipulation_selector" value="remove" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value="STRINGDOESNOTEXIST" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="miscellaneous" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="remove" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and trim to 4 inner-most bases --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="trim" /> - <param name="base_offset_type" value="offsets_absolute"/> - <param name="left_column_offset" value="45"/> - <param name="right_column_offset" value="45"/> - <param name="keep_zero_length" value="true" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="trim" /> + <conditional name="offset_type"> + <param name="base_offset_type" value="offsets_absolute"/> + <param name="left_column_offset" value="45"/> + <param name="right_column_offset" value="45"/> + </conditional> + <param name="keep_zero_length" value="true" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="fastq_trimmer_out1.fastqsanger" ftype="fastqsanger" /> </test> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="trim" /> - <param name="base_offset_type" value="offsets_percent"/> - <param name="left_column_offset" value="47.87"/> - <param name="right_column_offset" value="47.87"/> - <param name="keep_zero_length" value="true" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="trim" /> + <conditional name="offset_type"> + <param name="base_offset_type" value="offsets_percent"/> + <param name="left_column_offset" value="47.87"/> + <param name="right_column_offset" value="47.87"/> + </conditional> + <param name="keep_zero_length" value="true" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="fastq_trimmer_out1.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and rev comp --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rev_comp" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rev_comp" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_rev_comp.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and rev comp, with ambiguous DNA --> <test> <param name="input_file" value="misc_dna_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rev_comp" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rev_comp" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="misc_dna_as_sanger_rev_comp_1.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and rev comp, with ambiguous RNA --> <test> <param name="input_file" value="misc_rna_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rev_comp" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rev_comp" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="misc_rna_as_sanger_rev_comp_1.fastqsanger" ftype="fastqsanger" /> </test> <!-- match first seq and rev comp --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value="FAKE0001" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rev_comp" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value="FAKE0001" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rev_comp" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_rev_comp_1_seq.fastqsanger" ftype="fastqsanger" /> </test> <!-- match first seq and rev comp: i.e. undo above --> <test> <param name="input_file" value="sanger_full_range_rev_comp_1_seq.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value="FAKE0001" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rev_comp" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value="FAKE0001" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rev_comp" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and DNA to RNA --> <test> <param name="input_file" value="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="dna_to_rna" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="dna_to_rna" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_as_rna.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and RNA to DNA--> <test> <param name="input_file" value="sanger_full_range_as_rna.fastqsanger" ftype="fastqsanger" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rna_to_dna" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rna_to_dna" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger" /> </test> <!-- match all and RNA to DNA (gz compressed) --> <test> <param name="input_file" value="sanger_full_range_as_rna.fastqsanger.gz" ftype="fastqsanger.gz" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rna_to_dna" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rna_to_dna" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger.gz" decompress="true" /> </test> <!-- match all and RNA to DNA (bz2 compressed) --> <test> <param name="input_file" value="sanger_full_range_as_rna.fastqsanger.bz2" ftype="fastqsanger.bz2" /> - <param name="match_type_selector" value="identifier" /> - <param name="match_selector" value="regex" /> - <param name="match_by" value=".*" /> - <param name="manipulation_type_selector" value="sequence" /> - <param name="manipulation_selector" value="rna_to_dna" /> + <repeat name="match_blocks"> + <conditional name="match_type"> + <param name="match_type_selector" value="identifier" /> + <conditional name="match"> + <param name="match_selector" value="regex" /> + <param name="match_by" value=".*" /> + </conditional> + </conditional> + </repeat> + <repeat name="manipulate_blocks"> + <conditional name="manipulation_type"> + <param name="manipulation_type_selector" value="sequence" /> + <conditional name="manipulation"> + <param name="manipulation_selector" value="rna_to_dna" /> + </conditional> + </conditional> + </repeat> <output name="output_file" file="sanger_full_range_original_sanger.fastqsanger" ftype="fastqsanger.bz2" decompress="true" /> </test> </tests>