diff options
| -rw-r--r-- | bh20sequploader/bh20seq-options.yml | 17 | ||||
| -rw-r--r-- | bh20sequploader/bh20seq-schema.yml | 2 | ||||
| -rw-r--r-- | bh20simplewebuploader/main.py | 86 | ||||
| -rw-r--r-- | bh20simplewebuploader/templates/form.html | 19 | ||||
| -rw-r--r-- | scripts/from_genbank_to_fasta_and_yaml.py | 12 | ||||
| -rw-r--r-- | scripts/sequences.acc | 396 | ||||
| -rw-r--r-- | setup.py | 2 |
7 files changed, 336 insertions, 198 deletions
diff --git a/bh20sequploader/bh20seq-options.yml b/bh20sequploader/bh20seq-options.yml new file mode 100644 index 0000000..d05be5a --- /dev/null +++ b/bh20sequploader/bh20seq-options.yml @@ -0,0 +1,17 @@ +# Contains suggested human-readable field values and their corresponding IRIs. +# Keyed on the field names in the types in the schema. Relies on field names +# being unique or at least using the same options in different containing +# types. + +host_age_unit: + Years: http://purl.obolibrary.org/obo/UO_0000036 + Months: http://purl.obolibrary.org/obo/UO_0000035 + Weeks: http://purl.obolibrary.org/obo/UO_0000034 + Days: http://purl.obolibrary.org/obo/UO_0000033 + Hours: http://purl.obolibrary.org/obo/UO_0000032 + +host_sex: + Male: http://purl.obolibrary.org/obo/NCIT_C20197 + Female: http://purl.obolibrary.org/obo/NCIT_C27993 + Intersex: http://purl.obolibrary.org/obo/NCIT_C45908 + Unknown: http://purl.obolibrary.org/obo/NCIT_C17998 diff --git a/bh20sequploader/bh20seq-schema.yml b/bh20sequploader/bh20seq-schema.yml index 7ffc15b..4cd0865 100644 --- a/bh20sequploader/bh20seq-schema.yml +++ b/bh20sequploader/bh20seq-schema.yml @@ -30,7 +30,7 @@ $graph: # jsonldPredicate: # _id: http://purl.obolibrary.org/obo/NOMEN_0000037 host_sex: - doc: Sex of the host as define in NCIT, IRI expected (http://purl.obolibrary.org/obo/C20197 (Male), http://purl.obolibrary.org/obo/NCIT_C27993 (Female) or unkown (http://purl.obolibrary.org/obo/NCIT_C17998)) + doc: Sex of the host as defined in NCIT, IRI expected (http://purl.obolibrary.org/obo/NCIT_C20197 (Male), http://purl.obolibrary.org/obo/NCIT_C27993 (Female), http://purl.obolibrary.org/obo/NCIT_C45908 (Intersex), or http://purl.obolibrary.org/obo/NCIT_C17998 (Unknown)) type: string jsonldPredicate: _id: http://purl.obolibrary.org/obo/PATO_0000047 diff --git a/bh20simplewebuploader/main.py b/bh20simplewebuploader/main.py index f5324a5..8c5c18c 100644 --- a/bh20simplewebuploader/main.py +++ b/bh20simplewebuploader/main.py @@ -7,7 +7,7 @@ import sys import re import string import yaml -import urllib.request +import pkg_resources from flask import Flask, request, redirect, send_file, send_from_directory, render_template import os.path @@ -25,7 +25,7 @@ app.config['MAX_CONTENT_LENGTH'] = 50 * 1024 * 1024 @app.errorhandler(413) def handle_large_file(e): return (render_template('error.html', - error_message="One of your files is too large. The maximum file size is 1 megabyte."), 413) + error_message="One of your files is too large. The maximum file size is 50 megabytes."), 413) def type_to_heading(type_name): @@ -49,12 +49,32 @@ def name_to_label(field_name): return string.capwords(field_name.replace('_', ' ')) -def generate_form(schema): +def is_iri(string): """ - Linearize the schema and send a bunch of dicts. + Return True if the given string looks like an IRI, and False otherwise. + + Used for finding type IRIs in the schema. + + Right now only supports http(s) URLs because that's all we have in our schema. + """ + + return string.startswith('http') + +def generate_form(schema, options): + """ + Linearize the schema into a list of dicts. + Each dict either has a 'heading' (in which case we put a heading for a form section in the template) or an 'id', 'label', 'type', and 'required' - (in which case we make a form field in the template). + (in which case we make a form field in the template). Non-heading dicts + with type 'select' will have an 'options' field, with a list of (name, + value) tuples, and represent a form dropdown element. Non-heading dicts may + have a human-readable 'docstring' field describing them. + + Takes the deserialized metadata schema YAML, and also a deserialized YAML + of option values. The option values are keyed on (unscoped) field name in + the schema, and each is a dict of human readable option -> corresponding + IRI. """ # Get the list of form components, one of which is the root @@ -90,16 +110,35 @@ def generate_form(schema): for field_name, field_type in by_name.get(type_name, {}).get('fields', {}).items(): # For each field - ref_url = None + ref_iri = None + docstring = None if not isinstance(field_type, str): # If the type isn't a string + + # It may have documentation + docstring = field_type.get('doc', None) + # See if it has a more info/what goes here URL predicate = field_type.get('jsonldPredicate', {}) - if not isinstance(predicate, str): - ref_url = predicate.get('_id', None) + # Predicate may be a URL, a dict with a URL in _id, maybe a + # dict with a URL in _type, or a dict with _id and _type but no + # URLs anywhere. Some of these may not technically be allowed + # by the format, but if they occur, we might as well try to + # handle them. + if isinstance(predicate, str): + if is_iri(predicate): + ref_iri = predicate else: - ref_url = predicate # not sure this is correct - # Grab out its type field + # Assume it's a dict. Look at the fields we know about. + for field in ['_id', 'type']: + field_value = predicate.get(field, None) + if isinstance(field_value, str) and is_iri(field_value) and ref_iri is None: + # Take the first URL-looking thing we find + ref_iri = field_value + break + + + # Now overwrite the field type with the actual type string field_type = field_type.get('type', '') # Decide if the field is optional (type ends in ?) @@ -115,14 +154,26 @@ def generate_form(schema): for item in walk_fields(field_type, parent_keys + [field_name], subtree_optional or optional): yield item else: - # We know how to make a string input + # This is a leaf field. We need an input for it. record = {} record['id'] = '.'.join(parent_keys + [field_name]) record['label'] = name_to_label(field_name) record['required'] = not optional and not subtree_optional - if ref_url: - record['ref_url'] = ref_url - if field_type == 'string': + if ref_iri: + record['ref_iri'] = ref_iri + if docstring: + record['docstring'] = docstring + + if field_name in options: + # The field will be a 'select' type no matter what its real + # data type is. + record['type'] = 'select' # Not a real HTML input type. It's its own tag. + # We have a set of values to present + record['options'] = [] + for name, value in options[field_name].items(): + # Make a tuple for each one + record['options'].append((name, value)) + elif field_type == 'string': record['type'] = 'text' # HTML input type elif field_type == 'int': record['type'] = 'number' @@ -133,9 +184,10 @@ def generate_form(schema): return list(walk_fields(root_name)) -# At startup, we need to load the current metadata schema so we can make a form for it -METADATA_SCHEMA = yaml.safe_load(urllib.request.urlopen('https://raw.githubusercontent.com/arvados/bh20-seq-resource/master/bh20sequploader/bh20seq-schema.yml')) -FORM_ITEMS = generate_form(METADATA_SCHEMA) +# At startup, we need to load the metadata schema from the uploader module, so we can make a form for it +METADATA_SCHEMA = yaml.safe_load(pkg_resources.resource_stream("bh20sequploader", "bh20seq-schema.yml")) +METADATA_OPTION_DEFINITIONS = yaml.safe_load(pkg_resources.resource_stream("bh20sequploader", "bh20seq-options.yml")) +FORM_ITEMS = generate_form(METADATA_SCHEMA, METADATA_OPTION_DEFINITIONS) @app.route('/') def send_form(): diff --git a/bh20simplewebuploader/templates/form.html b/bh20simplewebuploader/templates/form.html index df66e8c..6993cf5 100644 --- a/bh20simplewebuploader/templates/form.html +++ b/bh20simplewebuploader/templates/form.html @@ -211,7 +211,6 @@ <div id="metadata_fill_form_spot"> <div id="metadata_fill_form"> - {{ record }} {% for record in fields %} {% if 'heading' in record %} @@ -221,15 +220,27 @@ <div class="record"> <h4>{{ record['heading'] }}</h4> {% else %} - <label for="{{ record['id'] }}"> + <label for="{{ record['id'] }}" title="{{ record.get('docstring', '') }}"> {{ record['label'] }} {{ "*" if record['required'] else "" }} - {% if 'ref_url' in record %} - <a href="{{ record['ref_url'] }}" title="More Info" target="_blank">?</a> + {% if 'docstring' in record %} + <a href='javascript:alert({{ record['docstring'] | tojson }})'>❓</a> + {% endif %} + {% if 'ref_iri' in record %} + <a href="{{ record['ref_iri'] }}" target="_blank" title="Ontology Link">🔗</a> {% endif %} </label> + {% if record['type'] == 'select' %} + <select id="{{ record['id'] }}" name="{{ record['id'] }}" {{ "required" if record['required'] else "" }}> + <option value="" selected>Choose one...</option> + {% for option in record['options'] %} + <option value="{{ option[1] }}">{{ option[0] }}</option> + {% endfor %} + </select> + {% else %} <input type="{{ record['type'] }}" id="{{ record['id'] }}" name="{{ record['id'] }}" {{ "required" if record['required'] else "" }}> {% endif %} + {% endif %} {% if loop.index == loop.length %} </div> {% endif %} diff --git a/scripts/from_genbank_to_fasta_and_yaml.py b/scripts/from_genbank_to_fasta_and_yaml.py index 0cc1a57..6a55b5e 100644 --- a/scripts/from_genbank_to_fasta_and_yaml.py +++ b/scripts/from_genbank_to_fasta_and_yaml.py @@ -7,7 +7,7 @@ import os path_ncbi_virus_accession = 'sequences.acc' -date = '20200414' +date = '20200415' path_seq_fasta = 'seq_from_nuccore.{}.fasta'.format(date) path_metadata_xml = 'metadata_from_nuccore.{}.xml'.format(date) @@ -19,9 +19,15 @@ for term in term_list: tmp_list = Entrez.read( Entrez.esearch(db='nuccore', term=term, idtype='acc', retmax='10000') )['IdList'] - print(term, len(tmp_list)) - + + # Remove mRNAs, ncRNAs, Proteins, and predicted models (more information here: https://en.wikipedia.org/wiki/RefSeq) + tmp_list = [x for x in tmp_list if x[:2] not in ['NM', 'NR', 'NP', 'XM', 'XR', 'XP', 'WP']] + # Remove the version in the id + tmp_list = [x.split('.')[0] for x in tmp_list] + + print(term, len(tmp_list)) + id_set.update([x.split('.')[0] for x in tmp_list]) print(term_list, len(id_set)) diff --git a/scripts/sequences.acc b/scripts/sequences.acc index 62bde2c..a420fb4 100644 --- a/scripts/sequences.acc +++ b/scripts/sequences.acc @@ -1,25 +1,159 @@ -MT325599 -MT325601 -MT325602 -MT325607 -MT325608 -MT325609 -MT325610 -MT325612 -MT325616 -MT325617 -MT325618 -MT325622 -MT325623 -MT325600 -MT325606 -MT325611 -MT325613 -MT325615 -MT325619 -MT325620 -MT325624 -MT325625 +NC_045512 +MT334522 +MT334523 +MT334524 +MT334525 +MT334526 +MT334527 +MT334528 +MT334529 +MT334530 +MT334531 +MT334532 +MT334533 +MT334534 +MT334535 +MT334536 +MT334537 +MT334538 +MT334539 +MT334540 +MT334541 +MT334542 +MT334543 +MT334544 +MT334545 +MT334546 +MT334555 +MT334547 +MT334548 +MT334549 +MT334550 +MT334551 +MT334552 +MT334553 +MT334554 +MT334556 +MT334557 +MT334558 +MT334559 +MT334560 +MT334561 +MT334562 +MT334563 +MT334564 +MT334565 +MT334566 +MT334567 +MT334568 +MT334569 +MT334570 +MT334571 +MT334572 +MT334573 +MT324062 +MT324680 +MT324684 +MT325573 +MT325574 +MT325576 +MT325577 +MT325578 +MT325580 +MT325591 +MT325592 +MT325593 +MT325595 +MT325605 +MT325627 +MT326028 +MT326029 +MT326031 +MT326048 +MT326093 +MT326092 +MT326091 +MT326090 +MT326085 +MT326084 +MT326083 +MT326082 +MT326081 +MT326080 +MT326077 +MT326067 +MT326057 +MT326024 +MT326025 +MT326032 +MT326033 +MT326035 +MT326036 +MT326037 +MT326040 +MT326041 +MT326043 +MT326044 +MT326046 +MT326049 +MT326050 +MT326052 +MT326053 +MT326055 +MT326056 +MT326059 +MT326062 +MT326063 +MT326066 +MT326069 +MT326070 +MT326071 +MT326073 +MT326074 +MT326075 +MT326088 +MT326089 +MT327745 +MT325568 +MT325572 +MT325575 +MT325583 +MT325584 +MT325604 +MT325631 +MT325632 +MT325635 +MT325636 +MT325637 +MT326095 +MT326096 +MT326103 +MT326112 +MT326113 +MT326114 +MT326115 +MT326122 +MT326131 +MT326132 +MT326133 +MT325563 +MT326164 +MT326166 +MT326167 +MT325570 +MT325579 +MT325581 +MT325582 +MT325586 +MT325594 +MT325598 +MT325626 +MT325628 +MT325633 +MT325634 +MT326030 +MT326038 +MT326058 MT325565 MT325566 MT326147 @@ -56,7 +190,6 @@ MT326121 MT326119 MT326109 MT326100 -MT325568 MT324679 MT325561 MT325571 @@ -95,31 +228,11 @@ MT326177 MT326184 MT326185 MT326187 -MT325572 -MT325575 -MT325583 -MT325584 -MT325604 -MT325631 -MT325632 -MT325635 -MT325636 -MT325637 -MT326095 -MT326096 -MT326103 -MT326112 -MT326113 -MT326114 -MT326115 -MT326122 -MT326131 -MT326132 -MT326133 -MT325563 -MT326164 -MT326166 -MT326167 +MT324681 +MT324682 +MT324683 +MT328032 +MT328035 MT325569 MT326097 MT326106 @@ -177,146 +290,83 @@ MT326101 MT326099 MT326098 MT326094 -MT326093 -MT326092 -MT326091 -MT326090 -MT326085 -MT326084 -MT326083 -MT326082 -MT326081 -MT326080 -MT326077 -MT326067 -MT326057 -MT326024 -MT326025 -MT326032 -MT326033 -MT326035 -MT326036 -MT326037 -MT326040 -MT326041 -MT326043 -MT326044 -MT326046 -MT326049 -MT326050 -MT326052 -MT326053 -MT326055 -MT326056 -MT326059 -MT326062 -MT326063 -MT326066 -MT326069 -MT326070 -MT326071 -MT326073 -MT326074 -MT326075 -MT326088 -MT326089 -MT327745 -MT324062 -MT324680 -MT324684 -MT325573 -MT325574 -MT325576 -MT325577 -MT325578 -MT325580 -MT325591 -MT325592 -MT325593 -MT325595 -MT325605 -MT325627 -MT326028 -MT326029 -MT326031 -MT326048 -MT325570 -MT325579 -MT325581 -MT325582 -MT325586 -MT325594 -MT325598 -MT325626 -MT325628 -MT325633 -MT325634 -MT326030 -MT326038 -MT326058 -MT324681 -MT324682 -MT324683 -MT328032 -MT328035 +MT325599 +MT325601 +MT325602 +MT325607 +MT325608 +MT325609 +MT325610 +MT325612 +MT325616 +MT325617 +MT325618 +MT325622 +MT325623 +MT325600 +MT325606 +MT325611 +MT325613 +MT325615 +MT325619 +MT325620 +MT325624 +MT325625 +MT322394 +MT322395 +MT322420 +MT322424 MT039874 MT077125 -MT322394 +MT322396 MT322397 -MT322398 MT322399 +MT322403 +MT322406 +MT322407 +MT322412 +MT322413 +MT322414 +MT322416 +MT322398 MT322400 MT322401 -MT322403 +MT322402 MT322404 MT322405 -MT322406 MT322408 MT322409 MT322410 MT322411 -MT322412 -MT322413 -MT322414 MT322415 -MT322416 MT322417 MT322418 MT322419 -MT322420 MT322421 MT322422 MT322423 -MT322424 -MT322396 -MT322402 -MT322395 -MT322407 -MT320538 MT320891 +MT320538 MT308692 MT308693 +MT308695 +MT308696 MT308698 MT308699 +MT308701 MT308703 MT308704 MT308694 -MT308695 -MT308696 MT308697 MT308700 -MT308701 MT308702 MT304476 MT304474 MT304475 MT293547 MT304477 -MT304483 -MT300186 MT304478 MT304479 -MT304480 MT304481 MT304482 MT304484 @@ -324,9 +374,12 @@ MT304485 MT304486 MT304487 MT304488 +MT304491 +MT304480 +MT304483 MT304489 MT304490 -MT304491 +MT300186 MT291831 MT291836 MT291834 @@ -366,7 +419,6 @@ MT293168 MT293175 MT293190 MT293191 -MT273658 MT293159 MT292582 MT293162 @@ -376,7 +428,6 @@ MT293165 MT293156 MT293157 MT293158 -MT281577 MT293171 MT293174 MT293176 @@ -426,6 +477,8 @@ MT293223 MT291826 MT291832 MT291833 +MT273658 +MT281577 MT281530 MT276331 MT276325 @@ -645,8 +698,6 @@ MT253700 MT253705 MT253709 MT253708 -MT233526 -MT246667 MT246451 MT246453 MT246454 @@ -689,6 +740,8 @@ MT246472 MT246473 MT246483 MT246484 +MT233526 +MT246667 MT240479 MT232869 MT232870 @@ -774,8 +827,8 @@ MT127113 MT127114 MT127115 MT126808 -LC528233 LC528232 +LC528233 MT123290 MT123291 MT123292 @@ -807,8 +860,8 @@ MT066159 MT066175 MT066176 LC523807 -LC523808 LC523809 +LC523808 MT044258 MT044257 MT042777 @@ -830,7 +883,6 @@ LC522350 MT027062 MT027063 MT027064 -MT020781 MT019530 MT019531 MT020881 @@ -838,13 +890,14 @@ MT019533 MT019529 MT019532 MT020880 +MT020781 LR757995 LR757996 LR757997 LR757998 MT007544 -MT008023 MT008022 +MT008023 MN996530 MN996531 MN996527 @@ -856,22 +909,21 @@ MN988668 MN988669 MN994467 MN988713 -MN938387 -MN938389 -MN975263 -MN975268 +MN938384 +MN975262 +MN985325 +MN975264 +MN975266 MN975267 +MN975268 MN938388 +MN938389 MN938390 -MN975264 -MN975265 -MN975266 -MN938386 +MN975263 MN938385 -MN938384 -MN975262 -MN985325 +MN938386 +MN938387 +MN975265 MN970003 MN970004 -NC_045512 MN908947 @@ -31,7 +31,7 @@ setup( author_email="peter.amstutz@curii.com", license="Apache 2.0", packages=["bh20sequploader", "bh20seqanalyzer", "bh20simplewebuploader"], - package_data={"bh20sequploader": ["bh20seq-schema.yml", "validation/formats"], + package_data={"bh20sequploader": ["bh20seq-schema.yml", "bh20seq-options.yml", "validation/formats"], }, install_requires=install_requires, extras_require={ |