Merge 38dc098 into 220522c

chanjo/calculate.py

@@ -30,3 +30,20 @@ def gene_metrics(self, *genes):
                      .filter(Transcript.gene_id.in_(genes))
                      .group_by(Transcript.gene_id))
         return query
+
+    def mean_cov_gene(self, gene_id, sample_id):
+        """Return all transcript stats for a gene"""
+        mean = None
+        result = (self.query(Transcript, TranscriptStat)
+                     .filter(Transcript.id == TranscriptStat.transcript_id)
+                     .filter(Transcript.gene_id==gene_id)
+                     .filter(TranscriptStat.sample_id == sample_id)
+                     .all()
+                     )
+        if not result:
+            return mean
+        mean = 0
+        for i,res in enumerate(result):
+            mean += res[1].mean_coverage
+        mean = mean/(i+1)
+        return mean

chanjo/cli/__init__.py

@@ -5,3 +5,4 @@
 from .sambamba import sambamba
 from .db import db_cmd
 from .init import init
+from .view import view

chanjo/cli/calculate.py

@@ -1,9 +1,11 @@
 # -*- coding: utf-8 -*-
 import json
 import logging
+from pprint import pprint as pp
 
 import click
 
+
 from chanjo.store.api import ChanjoDB
 from chanjo.store.constants import STAT_COLUMNS
 
@@ -37,3 +39,25 @@ 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('-i', '--gene-id', type=int, help='Gene id for a gene')
+@click.option('-s', '--sample', multiple=True, help='sample(s) to limit query to')
+@click.option('-p', '--pretty', is_flag=True)
+@click.pass_context
+def gene(context, gene_id, sample, pretty):
+    """Calculate average coverage for a gene"""
+    if not (gene_id or gene_symbol):
+        LOG.warning('Please specify a gene')
+        context.abort()
+    if not sample:
+        LOG.warning('Please specify at least one sample')
+        context.abort()
+
+    for sample_id in sample:
+        result = context.obj['db'].mean_cov_gene(gene_id, sample_id)
+        if result is None:
+            LOG.info("No result could be found")
+            return
+        row = {'average_coverage':result, 'sample_id': sample_id, 'gene': gene_id}
+        click.echo(dump_json(row, pretty=pretty))

chanjo/cli/view.py

@@ -0,0 +1,67 @@
+# -*- coding: utf-8 -*-
+import logging
+
+import click
+
+from chanjo.store.api import ChanjoDB
+from chanjo.store.models import TranscriptStat
+from .calculate import dump_json
+
+LOG = logging.getLogger(__name__)
+
+@click.group()
+@click.pass_context
+def view(context):
+    """View infromation from database."""
+    if not context.obj['database']:
+        LOG.warning("Please point to a database")
+        context.abort()
+    context.obj['db'] = ChanjoDB(uri=context.obj['database'])
+
+
+@view.command()
+@click.option('-s', '--sample', multiple=True, help='sample(s) to limit query to')
+@click.option('-g', '--group', multiple=True, help='group(s) to limit query to')
+@click.pass_context
+def sample(context, sample, group):
+    """View information on samples."""
+    result = context.obj['db'].sample(sample, group)
+    for res in result:
+        click.echo("sample_id: {0}, group_id: {1}, source: {2}".format(
+            res.id, res.group_id, res.source
+        ))
+
+@view.command()
+@click.option('-i', '--gene-id', type=int, help='Gene id for a gene')
+@click.option('--gene-symbol', help='Gene symbol for a gene')
+@click.option('-p', '--pretty', is_flag=True)
+@click.pass_context
+def gene(context, gene_id, gene_symbol, pretty):
+    """Display all transcripts for a gene"""
+    if not (gene_id or gene_symbol):
+        LOG.warning('Please specify a gene')
+        context.abort()
+
+    result = context.obj['db'].gene(gene_id, gene_symbol)
+    if result.count() == 0:
+        LOG.info("No genes found")
+
+    for res in result:
+        row = {
+            'id': res.id,
+            'gene_id': res.gene_id,
+            'gene_name': res.gene_name,
+            'chromosome': res.chromosome,
+            'length': res.length,
+        }
+        click.echo(dump_json(row, pretty=pretty))
+
+@view.command()
+@click.option('-s', '--sample', help='sample to limit query to')
+@click.pass_context
+def incomplete(context, sample):
+    """Wiew all transcripts with incomplete exons for a sample"""
+    query = context.obj['db'].query(TranscriptStat).filter_by(sample_id=sample)
+    result = query.filter(TranscriptStat._incomplete_exons.isnot(None))
+    for res in result:
+        print(res)

chanjo/store/api.py

@@ -6,7 +6,7 @@
 from alchy import Manager
 
 from chanjo.calculate import CalculateMixin
-from .models import BASE
+from .models import (BASE, Sample, Transcript)
 
 log = logging.getLogger(__name__)
 
@@ -122,3 +122,22 @@ def save(self):
             self.session.rollback()
             raise error
         return self
+
+    def sample(self, sample_ids=None, groups=None):
+        """Return samples from database"""
+        sql_query = self.query(Sample)
+        if sample_ids:
+            sql_query = sql_query.filter(Sample.id.in_(sample_ids))
+        if groups:
+            sql_query = sql_query.filter(Sample.group_id.in_(groups))
+
+        return sql_query
+
+    def gene(self, gene_id=None, gene_symbol=None):
+        """Return transcripts from certain genes"""
+        sql_query = self.query(Transcript)
+        if gene_id:
+            sql_query = sql_query.filter_by(gene_id=gene_id)
+        if gene_symbol:
+            sql_query = sql_query.filter_by(gene_name=gene_symbol)
+        return sql_query

setup.py

@@ -101,6 +101,7 @@ def run_tests(self):
             'load = chanjo.cli:load',
             'link = chanjo.cli:link',
             'calculate = chanjo.cli:calculate',
+            'view = chanjo.cli:view',
         ]
     },