Skip to content

[Question] should n and nm be supported for placements?  #35

Open
@antgonza

Description

@antgonza

While running some tests with the SEPP placements in Qiita we noticed that the current code rewrites the placements as:

plcmnts['placements'].extend([{'p': placement, 'nm': [[sequence, 1]]}
                           for sequence, placement in placements.items()])

which is currently not supported by bp so we had to rewrite them as:

plcmnts['placements'].extend([{'p': placement, 'n': [sequence, ]}
                           for sequence, placement in placements.items()])

Thus, wondering if this is enough or if there is any reason why bp should support both.

cc: @sjanssen2; BTW the current version can add all the Qiita 150bp deblur fragments into the GG/SEPP backbone in < 3.5hrs and ~200Gb.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions