geoseeq 0.6.8a2__tar.gz → 0.6.9__tar.gz

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geoseeq
-Version: 0.6.8a2
+Version: 0.6.9
 Summary: GeoSeeq command line tools and python API
 Author: David C. Danko
 Author-email: "David C. Danko" <dcdanko@biotia.io>

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/geoseeq/cli/download.py

@@ -98,6 +98,7 @@ def cli_download_metadata(state, sample_ids):
 
 cores_option = click.option('--cores', default=1, help='Number of downloads to run in parallel')
 head_option = click.option('--head', default=None, type=int, help='Download the first N bytes of each file')
+alt_id_option = click.option('--alt-sample-id', default=None, help='Specify an alternate sample id from the project metadata to id samples')
 
 @cli_download.command("files")
 @use_common_state
@@ -113,6 +114,7 @@ head_option = click.option('--head', default=None, type=int, help='Download the
 @click.option("--extension", multiple=True, help="Only download files with this extension. e.g. 'fastq.gz', 'bam', 'csv'")
 @click.option("--with-versions/--without-versions", default=False, help="Download all versions of a file, not just the latest")
 @ignore_errors_option
+@alt_id_option
 @project_id_arg
 @sample_ids_arg
 def cli_download_files(
@@ -129,6 +131,7 @@ def cli_download_files(
     with_versions,
     download,
     ignore_errors,
+    alt_sample_id,
     project_id,
     sample_ids,
 ):
@@ -164,6 +167,13 @@ def cli_download_files(
         haib17CEM4890_H2NYMCCXY_SL254769 haib17CEM4890_H2NYMCCXY_SL254773 `# specify the samples by name` \\ 
         --folder-type sample --extension '.contigs.fasta' # filter for contig files
 
+    \b
+    # Download files from a sample in the metasub project using an alternate sample id called "barcode"
+    $ geoseeq download files 'MetaSUB Consortium/Cell Paper' `# specify the project` \\
+        235183938 `# the alternate sample name (in this case a barcode number)` \\
+        --alt-sample-id barcode `# specify the alternate sample id column name` \\
+        --folder-type 'sample' `# only download files from sample folders`
+
     ---
 
     Command Arguments:
@@ -184,7 +194,7 @@ def cli_download_files(
     samples = []
     if sample_ids:
         logger.info(f"Fetching info for {len(sample_ids)} samples.")
-        samples = handle_multiple_sample_ids(knex, sample_ids, proj=proj)
+        samples = handle_multiple_sample_ids(knex, sample_ids, proj=proj, alternate_id_col=alt_sample_id)
 
     response = proj.bulk_find_files(
         sample_uuids=[s.uuid for s in samples],
@@ -377,9 +387,21 @@ def cli_download_ids(state, cores, target_dir, file_name, yes, download, head, i
 @click.option("--download/--urls-only", default=True, help="Download files or just print urls")
 @module_option(FASTQ_MODULE_NAMES, use_default=False)
 @ignore_errors_option
+@alt_id_option
 @project_id_arg
 @sample_ids_arg
-def cli_download_fastqs(state, cores, target_dir, yes, first, download, module_name, ignore_errors, project_id, sample_ids):
+def cli_download_fastqs(state,
+                        cores,
+                        target_dir,
+                        yes,
+                        first,
+                        download,
+                        module_name,
+                        ignore_errors,
+                        alt_sample_id,
+                        project_id,
+                        sample_ids
+):
     """Download fastq files from a GeoSeeq project.
 
     This command will download fastq files from a GeoSeeq project. You can filter
@@ -401,6 +423,10 @@ def cli_download_fastqs(state, cores, target_dir, yes, first, download, module_n
     # Download all fastq files from two samples in "My Org/My Project"
     $ geoseeq download fastqs "My Org/My Project" S1 S2
 
+    \b
+    # Download all fastq files from a single sample using an alternate sample id called "barcode"
+    $ geoseeq download fastqs 'MetaSUB Consortium/Cell Paper' 235183938 --alt-sample-id barcode
+
     ---
 
     Command Arguments:
@@ -422,7 +448,7 @@ def cli_download_fastqs(state, cores, target_dir, yes, first, download, module_n
     samples = []
     if sample_ids:
         logger.info(f"Fetching info for {len(sample_ids)} samples.")
-        samples = handle_multiple_sample_ids(knex, sample_ids, proj=proj)
+        samples = handle_multiple_sample_ids(knex, sample_ids, proj=proj, alternate_id_col=alt_sample_id)
     else:
         logger.info("Fetching info for all samples in project.")
         samples = proj.get_samples()

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/geoseeq/cli/main.py

@@ -54,7 +54,7 @@ def version():
     Use of this tool implies acceptance of the GeoSeeq End User License Agreement.
     Run `geoseeq eula show` to view the EULA.
     """
-    click.echo('0.6.8a2')  # remember to update setup
+    click.echo('0.6.9')  # remember to update pyproject.toml
 
 
 @main.group('advanced')

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/geoseeq/cli/shared_params/id_handlers.py

@@ -133,7 +133,30 @@ def handle_folder_id(knex, folder_id, yes=False, private=True, create=True):
     raise ValueError('sample_folder_id must be a UUID, an organization name and project name, or a GRN')
 
 
-def handle_multiple_sample_ids(knex, sample_ids, proj=None):
+def map_alternate_ids_to_uuids(proj, alternate_id_col, sample_ids):
+    """Return a list of sample UUIDs
+    
+    `proj` is a project object
+    `alternate_id_col` is the name of the column containing alternate IDs
+    `sample_ids` is a list of alternate IDs
+    """
+    metadata = proj.get_sample_metadata()
+    if alternate_id_col not in metadata:
+        raise ValueError(f'Column "{alternate_id_col}" not found in project metadata')
+    alt_col_df = metadata[["uuid", alternate_id_col]]
+    # filter to the alt ids in our list- it is possible alt_id_col as a whole is not
+    # unique but that our list of alt ids is
+    alt_col_df = alt_col_df[alt_col_df[alternate_id_col].isin(sample_ids)]
+    if alt_col_df.shape[0] == 0:
+        raise ValueError(f'No samples found with the given alternate IDs in list')
+    if alt_col_df.shape[0] < len(sample_ids):
+        raise ValueError(f'Not all alternate IDs in list are found')
+    if alt_col_df.shape[0] > len(sample_ids):
+        raise ValueError(f'More than one sample found with the same alternate ID')
+    return list(alt_col_df['uuid'])
+
+
+def handle_multiple_sample_ids(knex, sample_ids, proj=None, alternate_id_col=None):
     """Return a list of fetched sample objects
     
     `sample_ids` may have three different structures:
@@ -144,7 +167,9 @@ def handle_multiple_sample_ids(knex, sample_ids, proj=None):
     Any sample may in fact be a file containing sample IDs, in which case the file will be read line by line
     and each element will be a sample ID
 
-    If `one_project` is True, all samples must be from the same project
+    If `proj` is provided then `alternate_id_col` may also be provided.
+    If so then alternate IDs will be used to fetch samples. If alternate ids are 
+    not present or not unique then fail.
     """
     project_as_arg = bool(proj)
     if proj or (proj := el_is_project_id(knex, sample_ids[0])):
@@ -155,7 +180,10 @@ def handle_multiple_sample_ids(knex, sample_ids, proj=None):
             return list(proj.get_samples(cache=False))
         else:
             samples = []
-            for el in flatten_list_of_els_and_files(sample_ids):
+            sample_ids = flatten_list_of_els_and_files(sample_ids)
+            if alternate_id_col:
+                sample_ids = map_alternate_ids_to_uuids(proj, alternate_id_col, sample_ids)
+            for el in sample_ids:
                 if is_grn_or_uuid(el):
                     el = el.split(':')[-1]
                     samples.append(sample_from_uuid(knex, el))

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/geoseeq/knex.py

@@ -214,7 +214,8 @@ class Knex:
         """Return a knex authenticated with a profile."""
         endpoint, token = load_auth_profile(profile)
         knex = cls(endpoint)
-        knex.add_api_token(token)
+        if token:
+            knex.add_api_token(token)
         return knex
     
 

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/geoseeq/project.py

@@ -256,9 +256,14 @@ class Project(RemoteObject):
 
     def get_sample_metadata(self):
         """Return a pandas dataframe with sample metadata."""
-        url = f"sample_groups/{self.uuid}/metadata"
-        blob = self.knex.get(url)
-        return pd.DataFrame.from_dict(blob, orient="index")
+        url = f"sample_groups/{self.uuid}/samples-list?page=1&page_size=500&&"
+        rows = []
+        while url:
+            blob = self.knex.get(url)
+            rows.extend(blob["results"])
+            url = blob["next"]
+        return pd.DataFrame(rows)
+    
     
     @property
     def n_samples(self):

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/geoseeq/utils.py

@@ -16,11 +16,19 @@ logger.addHandler(logging.NullHandler())  # No output unless configured by calli
 def load_auth_profile(profile=""):
     """Return an endpoit and a token"""
     profile = profile or "__default__"
-    with open(PROFILES_PATH, "r") as f:
-        profiles = json.load(f)
-    if profile in profiles:
-        return profiles[profile]["endpoint"], profiles[profile]["token"]
-    raise KeyError(f"Profile {profile} not found.")
+    try:
+        with open(PROFILES_PATH, "r") as f:
+            profiles = json.load(f)
+        if profile in profiles:
+            return profiles[profile]["endpoint"], profiles[profile]["token"]
+        raise KeyError(f"Profile {profile} not found.")
+    except FileNotFoundError:
+        endpoint, token = environ.get("GEOSEEQ_ENDPOINT", DEFAULT_ENDPOINT), environ.get("GEOSEEQ_API_TOKEN", None)
+        if token:
+            logger.debug("Using environment variables for authentication.")
+        else:
+            logger.warning("Accessing anonymously, functionality may be limited. Configure profiles or set GEOSEEQ_API_TOKEN to authenticate.")
+        return endpoint, token
 
 
 def set_profile(token, endpoint=DEFAULT_ENDPOINT, profile="", overwrite=False):

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/geoseeq.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geoseeq
-Version: 0.6.8a2
+Version: 0.6.9
 Summary: GeoSeeq command line tools and python API
 Author: David C. Danko
 Author-email: "David C. Danko" <dcdanko@biotia.io>

{geoseeq-0.6.8a2 → geoseeq-0.6.9}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "geoseeq"
-version = "0.6.8a2"
+version = "0.6.9"
 authors = [
   { name="David C. Danko", email="dcdanko@biotia.io" },
 ]