Expand CLI

.travis.yml

@@ -13,6 +13,7 @@ install:
     - mv sambamba_v0.5.8 ./bin/sambamba
     - chmod +x ./bin/sambamba
     - export PATH="${PATH}:${PWD}/bin"
+    - pip install six==1.12.0
     - pip install -r requirements.txt -r requirements-dev.txt .
 
 before_script:

CHANGELOG.md

@@ -2,6 +2,11 @@
 All notable changes to this project will be documented in this file.
 This project adheres to [Semantic Versioning](http://semver.org/).
 
+## Unreleased
+
+### Added
+- added CLI commands
+
 ## [4.0.0] - 2016-08-02
 Version 4 slims down Chanjo quite a bit.
 

chanjo/calculate.py

@@ -1,7 +1,8 @@
-# -*- coding: utf-8 -*-
+"""Module for calculate operations on database"""
+
 from sqlalchemy.sql import func
 
-from chanjo.store.models import Sample, Transcript, TranscriptStat
+from chanjo.store.models import Transcript, TranscriptStat
 
 
 class CalculateMixin:
@@ -10,23 +11,48 @@ class CalculateMixin:
 
     def mean(self, sample_ids=None):
         """Calculate the mean values of all metrics per sample."""
-        sql_query = (self.query(TranscriptStat.sample_id,
-                                func.avg(TranscriptStat.mean_coverage),
-                                func.avg(TranscriptStat.completeness_10),
-                                func.avg(TranscriptStat.completeness_15),
-                                func.avg(TranscriptStat.completeness_20),
-                                func.avg(TranscriptStat.completeness_50),
-                                func.avg(TranscriptStat.completeness_100))
-                         .group_by(TranscriptStat.sample_id))
+        sql_query = self.query(
+            TranscriptStat.sample_id,
+            func.avg(TranscriptStat.mean_coverage),
+            func.avg(TranscriptStat.completeness_10),
+            func.avg(TranscriptStat.completeness_15),
+            func.avg(TranscriptStat.completeness_20),
+            func.avg(TranscriptStat.completeness_50),
+            func.avg(TranscriptStat.completeness_100),
+        ).group_by(TranscriptStat.sample_id)
         if sample_ids:
             sql_query = sql_query.filter(TranscriptStat.sample_id.in_(sample_ids))
         return sql_query
 
     def gene_metrics(self, *genes):
         """Calculate gene statistics."""
-        query = (self.mean()
-                     .add_column(Transcript.gene_id)
-                     .join(TranscriptStat.transcript)
-                     .filter(Transcript.gene_id.in_(genes))
-                     .group_by(Transcript.gene_id))
+        query = (
+            self.mean()
+            .add_column(Transcript.gene_id)
+            .join(TranscriptStat.transcript)
+            .filter(Transcript.gene_id.in_(genes))
+            .group_by(Transcript.gene_id)
+        )
         return query
+
+    def sample_coverage(self, sample_ids: list, genes: list) -> dict:
+        """Calculate coverage for samples."""
+        query = self.query(
+            TranscriptStat.sample_id.label('sample_id'),
+            func.avg(TranscriptStat.mean_coverage).label('mean_coverage'),
+            func.avg(TranscriptStat.completeness_10).label('mean_completeness'),
+        ).join(
+            Transcript,
+        ).filter(
+            Transcript.gene_id.in_(genes),
+            TranscriptStat.sample_id.in_(sample_ids),
+        ).group_by(TranscriptStat.sample_id)
+
+        data = {
+            result.sample_id: {
+                "mean_coverage": result.mean_coverage,
+                "mean_completeness": result.mean_completeness,
+            }
+            for result in query
+        }
+        return data

chanjo/cli/calculate.py

@@ -1,11 +1,11 @@
-# -*- coding: utf-8 -*-
+"""Functions for calculating operations on database"""
 import json
 import logging
 
 import click
 
 from chanjo.store.api import ChanjoDB
-from chanjo.store.constants import STAT_COLUMNS
+from chanjo.store.constants import STAT_COLUMNS, OMIM_GENE_IDS
 
 LOG = logging.getLogger(__name__)
 
@@ -37,3 +37,17 @@ def mean(context, sample, pretty):
     for result in query:
         row = {column: value for column, value in zip(columns, result)}
         click.echo(dump_json(row, pretty=pretty))
+
+
+@calculate.command()
+@click.option('-p', '--pretty', is_flag=True, help="Print in pretty format")
+@click.option('-s', '--sample', multiple=True, type=str, help="Sample to get coverage for")
+@click.option('-o', '--omim', is_flag=True, help="Use genes in the OMIM panel")
+@click.argument('genes', nargs=-1)
+@click.pass_context
+def coverage(context, pretty, sample, omim, genes):
+    """Calculate coverage for sample on specified genes"""
+    if omim:
+        genes = OMIM_GENE_IDS
+    query = context.obj['db'].sample_coverage(sample_ids=sample, genes=list(genes))
+    click.echo(dump_json(query, pretty=pretty))

chanjo/cli/db.py

@@ -1,4 +1,6 @@
-# -*- coding: utf-8 -*-
+"""CLI commands to interact with db"""
+
+import json
 import logging
 
 import click
@@ -9,36 +11,83 @@
 LOG = logging.getLogger(__name__)
 
 
-@click.group('db')
+@click.group("db")
 @click.pass_context
 def db_cmd(context):
     """Interact with the database for maintainance tasks."""
-    context.obj['db'] = ChanjoDB(uri=context.obj['database'])
+    context.obj["db"] = ChanjoDB(uri=context.obj["database"])
 
 
 @db_cmd.command()
-@click.option('--reset', is_flag=True, help='tear down existing db')
+@click.option("--reset", is_flag=True, help="tear down existing db")
 @click.pass_context
 def setup(context, reset):
     """Initialize a new datbase from scratch."""
     if reset:
-        LOG.info('tearing down existing database')
-        context.obj['db'].tear_down()
-    LOG.info('setting up new database')
-    context.obj['db'].set_up()
+        LOG.info("tearing down existing database")
+        context.obj["db"].tear_down()
+    LOG.info("setting up new database")
+    context.obj["db"].set_up()
 
 
 @db_cmd.command()
-@click.argument('sample_id', type=str)
+@click.argument("sample_id", type=str)
 @click.pass_context
 def remove(context, sample_id):
     """Remove all traces of a sample from the database."""
-    store = context.obj['db']
-    LOG.debug('find sample in database with id: %s', sample_id)
+    store = context.obj["db"]
+    LOG.debug("find sample in database with id: %s", sample_id)
     sample_obj = Sample.query.get(sample_id)
     if sample_obj is None:
-        LOG.warning('sample (%s) not found in database', sample_id)
+        LOG.warning("sample (%s) not found in database", sample_id)
         context.abort()
-    LOG.info('delete sample (%s) from database', sample_id)
+    LOG.info("delete sample (%s) from database", sample_id)
     store.session.delete(sample_obj)
     store.save()
+
+
+@db_cmd.command()
+@click.option("--group-id", "-g", help="Display samples in a group")
+@click.option("--sample-id", "-s", help="Display sample")
+@click.option("--pretty", "-p", is_flag=True, help="Print in pretty format")
+@click.pass_context
+def samples(context, group_id, sample_id, pretty):
+    """Display samples from database"""
+
+    store = context.obj["db"]
+    query = store.fetch_samples(sample_id=sample_id, group_id=group_id)
+    indent = None
+    if pretty:
+        indent = 4
+    click.echo(
+        json.dumps([dict(result) for result in query], default=str, indent=indent)
+    )
+
+
+@db_cmd.command()
+@click.option("--sample-id", "-s", help="Samples to limit query to")
+@click.option("--pretty", "-p", is_flag=True, help="Print in pretty format")
+@click.pass_context
+def transcripts(context, sample_id, pretty):
+    """Display transcripts from database"""
+    store = context.obj["db"]
+    query = store.fetch_transcripts(sample_id=sample_id)
+    indent = None
+    if pretty:
+        indent = 4
+    click.echo(
+        json.dumps([dict(result) for result in query], default=str, indent=indent)
+    )
+
+
+@db_cmd.command()
+@click.option("--group-id", "-g", help="Delete entire group")
+@click.option("--sample-id", "-s", help="Delete sample")
+@click.pass_context
+def delete(context, group_id, sample_id):
+    """Delete from database"""
+    store = context.obj["db"]
+    if sample_id:
+        store.delete_sample(sample_id=sample_id)
+    elif group_id:
+        store.delete_group(group_id=group_id)

chanjo/store/api.py

@@ -4,14 +4,15 @@
 import os
 
 from alchy import Manager
-
 from chanjo.calculate import CalculateMixin
 from .models import BASE
+from .fetch import FetchMixin
+from .delete import DeleteMixin
 
 log = logging.getLogger(__name__)
 
 
-class ChanjoDB(Manager, CalculateMixin):
+class ChanjoDB(Manager, CalculateMixin, DeleteMixin, FetchMixin):
     """SQLAlchemy-based database object.
 
     Bundles functionality required to setup and interact with various