Provides files and formats for handling structure data.
Reading structure files
Structured sequence files contain intra-molecular interactions of RNA or protein. Usually, but not necessarily, they contain the nucleotide or amino acid sequences and descriptions as well. Interactions can be represented either as fixed secondary structure, where every character is assigned at most one interaction partner (structure of minimum free energy), or an annotated sequence, where every character is assigned a set of interaction partners with specific base pair probabilities.
The structured sequence file abstraction supports reading ten different fields:
The first three fields are retrieved by default (and in that order). The seqan3::field::structured_seq may be selected to have sequence and structure directly stored in a more memory-efficient combined container. If you select this field you must not select seqan3::field::seq or seqan3::field::structure.
Construction and specialisation
This class comes with two constructors, one for construction from a file name and one for construction from an existing stream and a known format. The first one automatically picks the format based on the extension of the file name. The second can be used if you have a non-file stream, like std::cin or std::istringstream, that you want to read from and/or if you cannot use file-extension based detection, but know that your input file has a certain format.
In most cases the template parameters are deduced completely automatically, e.g. reading from a std::istringstream:
Note that this is not the same as writing structure_file_input<> (with angle brackets). In the latter case they are explicitly set to their default values, in the former case automatic deduction happens which chooses different parameters depending on the constructor arguments. For opening from file, structure_file_input<> would have also worked, but for opening from stream it would not have.
In some cases, you do need to specify the arguments, e.g. if you want to read amino acids:
You can define your own traits type to further customise the types used by and returned by this class, see seqan3::structure_file_default_traits_rna for more details. As mentioned above, specifying at least one template parameter yourself means that you loose automatic deduction so if you want to read amino acids and want to read from a string stream you need to give all types yourself:
In the above example, rec has the type seqan3::structure_file_input::record_type which is a specialisation of seqan3::record and behaves like a std::tuple (that's why we can access it via get). Instead of using the seqan3::field based interface on the record, you could also use std::get<0> or even std::get<rna4_vector> to retrieve the sequence, but it is not recommended, because it is more error-prone.
Note: It is important to write auto & and not just auto, otherwise you will copy the record on every iteration. Since the buffer gets "refilled" on every iteration, you can also move the data out of the record if you want to store it somewhere without copying:
If you want to skip specific fields from the record you can pass a non-empty fields trait object to the structure_file_input constructor to select the fields that should be read from the input. For example to choose a combined field for SEQ and STRUCTURE (see above). Or to never actually read the STRUCTURE, if you don't need it. The following snippets demonstrate the usage of such a fields trait object.
When reading a file, all fields not present in the file (but requested implicitly or via the selected_field_ids parameter) are ignored.
Views on files
Since SeqAn files are ranges, you can also create views over files. A useful example is to filter the records based on certain criteria, e.g. minimum length of the sequence field:
You can check whether a file is at end by comparing begin() and end() (if they are the same, the file is at end).
Formats
Currently, the only implemented format is seqan3::format_vienna. More formats will follow soon.
Writing structure files
Structured sequence files contain intra-molecular interactions of RNA or protein. Usually, but not necessarily, they contain the nucleotide or amino acid sequences and descriptions as well. Interactions can be represented either as fixed secondary structure, where every character is assigned at most one interaction partner (structure of minimum free energy), or an annotated sequence, where every character is assigned a set of interaction partners with specific base pair probabilities.
The structured sequence file abstraction supports writing ten different fields:
The member functions take any and either of these fields. If the field ID of an argument cannot be deduced, it is assumed to correspond to the field ID of the respective template parameter.
Construction and specialisation
This class comes with two constructors, one for construction from a file name and one for construction from an existing stream and a known format. The first one automatically picks the format based on the extension of the file name. The second can be used if you have a non-file stream, like std::cout or std::ostringstream, that you want to read from and/or if you cannot use file-extension based detection, but know that your output file has a certain format.
In most cases the template parameters are deduced completely automatically:
Note that this is not the same as writing structure_file_output<> (with angle brackets). In the latter case they are explicitly set to their default values, in the former case automatic deduction happens which chooses different parameters depending on the constructor arguments. Prefer deduction over explicit defaults.
The easiest way to write to a sequence file is to use the push_back() or emplace_back() member functions. These work similarly to how they work on a std::vector. If you pass a tuple to push_back() or give arguments to emplace_back() the seqan3::field ID of the i-th tuple-element/argument is assumed to be the i-th value of selected_field_ids, i.e. by default the first is assumed to be seqan3::field::seq, the second seqan3::field::id and the third one seqan3::field::structure. You may give less fields than are selected, if the actual format you are writing to can cope with less (e.g. for Vienna it is sufficient to write seqan3::field::seq, seqan3::field::id and seqan3::field::structure, even if selected_field_ids also contains seqan3::field::energy).
You may also use the output file's iterator for writing, however, this rarely provides an advantage.
Writing record-wise (custom fields)
If you want to pass a combined object for SEQ and STRUCTURE fields to push_back() / emplace_back(), or if you want to change the order of the parameters, you can pass a non-empty fields trait object to the structure_file_output constructor to select the fields that are used for interpreting the arguments.
The following snippets demonstrates the usage of such a fields trait object.
Provides the composite of nucleotide with structure alphabets.
A different way of passing custom fields to the file is to pass a seqan3::record – instead of a tuple – to push_back(). The seqan3::record clearly indicates which of its elements has which seqan3::field ID so the file will use that information instead of the template argument. This is especially handy when reading from one file and writing to another, because you don't have to configure the output file to match the input file, it will just work:
{"GATA"_rna5, "Third", "...."_wuss51}}; // a range of "records"
fout = range; // will iterate over the records and write them
}
File I/O pipelines
Record-wise writing in batches also works for writing from input files directly to output files, because input files are also input ranges in SeqAn. This can be combined with file-based views to create I/O pipelines:
The record-based interface treats the file as a range of tuples (the records), but in certain situations you might have the data as columns, i.e. a tuple-of-ranges, instead of range-of-tuples.
You can use column-based writing in that case, it uses operator=() and seqan3::views::zip():
Collaboration diagram for Structure File:
1.9.6