[galaxy-commits] commit/galaxy-central: jgoecks: Trackster: improve method for fetching additional data and simply data fetching.
Bitbucket
commits-noreply at bitbucket.org
Tue Sep 13 16:55:26 EDT 2011
1 new changeset in galaxy-central:
http://bitbucket.org/galaxy/galaxy-central/changeset/c6452f6558ee/
changeset: c6452f6558ee
user: jgoecks
date: 2011-09-13 22:55:17
summary: Trackster: improve method for fetching additional data and simply data fetching.
affected #: 3 files (1.4 KB)
--- a/lib/galaxy/visualization/tracks/data_providers.py Mon Sep 12 14:51:26 2011 -0400
+++ b/lib/galaxy/visualization/tracks/data_providers.py Tue Sep 13 16:55:17 2011 -0400
@@ -3,7 +3,7 @@
"""
import sys
-from math import floor, ceil, log, pow
+from math import ceil, log
import pkg_resources
pkg_resources.require( "bx-python" )
if sys.version_info[:2] == (2, 4):
@@ -13,14 +13,12 @@
from galaxy.datatypes.util.gff_util import *
from galaxy.util.json import from_json_string
from bx.interval_index_file import Indexes
-from bx.arrays.array_tree import FileArrayTreeDict
from bx.bbi.bigwig_file import BigWigFile
from galaxy.util.lrucache import LRUCache
from galaxy.visualization.tracks.summary import *
import galaxy_utils.sequence.vcf
from galaxy.datatypes.tabular import Vcf
from galaxy.datatypes.interval import Bed, Gff, Gtf
-from galaxy.datatypes.util.gff_util import parse_gff_attributes
from pysam import csamtools, ctabix
@@ -32,7 +30,20 @@
return None
else:
return float(n)
-
+
+def get_bounds( reads, start_pos_index, end_pos_index ):
+ """
+ Returns the minimum and maximum position for a set of reads.
+ """
+ max_low = sys.maxint
+ max_high = -sys.maxint
+ for read in reads:
+ if read[ start_pos_index ] < max_low:
+ max_low = read[ start_pos_index ]
+ if read[ end_pos_index ] > max_high:
+ max_high = read[ end_pos_index ]
+ return max_low, max_high
+
class TracksDataProvider( object ):
""" Base class for tracks data providers. """
@@ -75,8 +86,11 @@
def get_data( self, chrom, start, end, start_val=0, max_vals=None, **kwargs ):
"""
Returns data in region defined by chrom, start, and end. start_val and
- max_vals are used to denote the data to return: start_val is the first value to
+ max_vals are used to denote the data to return: start_val is the first element to
return and max_vals indicates the number of values to return.
+
+ Return value must be a dictionary with the following attributes:
+ dataset_type, data
"""
# Override.
pass
@@ -218,18 +232,23 @@
def get_data( self, chrom, start, end, start_val=0, max_vals=sys.maxint, **kwargs ):
"""
- Fetch reads in the region.
+ Fetch reads in the region and additional metadata.
- Each read is a list with the format
- [<guid>, <start>, <end>, <name>, <read_1>, <read_2>]
- where <read_1> has the format
- [<start>, <end>, <cigar>, ?<read_seq>?]
- and <read_2> has the format
- [<start>, <end>, <cigar>, ?<read_seq>?]
- For single-end reads, read has format:
- [<guid>, <start>, <end>, <name>, cigar, seq]
- NOTE: read end and sequence data are not valid for reads outside of
- requested region and should not be used.
+ Returns a dict with the following attributes:
+ data - a list of reads with the format
+ [<guid>, <start>, <end>, <name>, <read_1>, <read_2>]
+ where <read_1> has the format
+ [<start>, <end>, <cigar>, ?<read_seq>?]
+ and <read_2> has the format
+ [<start>, <end>, <cigar>, ?<read_seq>?]
+ For single-end reads, read has format:
+ [<guid>, <start>, <end>, <name>, cigar, seq]
+ NOTE: read end and sequence data are not valid for reads outside of
+ requested region and should not be used.
+
+ max_low - lowest coordinate for the returned reads
+ max_high - highest coordinate for the returned reads
+ message - error/informative message
"""
start, end = int(start), int(end)
orig_data_filename = self.original_dataset.file_name
@@ -304,9 +323,13 @@
r2 = [ read['mate_start'], read['mate_start'] ]
results.append( [ "%i_%s" % ( read_start, qname ), read_start, read_end, qname, r1, r2 ] )
-
+
+ # Clean up.
bamfile.close()
- return { 'data': results, 'message': message }
+
+ max_low, max_high = get_bounds( results, 1, 2 )
+
+ return { 'data': results, 'message': message, 'max_low': max_low, 'max_high': max_high }
class BBIDataProvider( TracksDataProvider ):
"""
@@ -334,9 +357,10 @@
return None
all_dat = all_dat[0] # only 1 summary
- return { 'max': float( all_dat['max'] ), \
- 'min': float( all_dat['min'] ), \
- 'total_frequency': float( all_dat['coverage'] ) }
+ return { 'data' : { 'max': float( all_dat['max'] ), \
+ 'min': float( all_dat['min'] ), \
+ 'total_frequency': float( all_dat['coverage'] ) } \
+ }
start = int(start)
end = int(end)
@@ -361,7 +385,7 @@
result.append( (pos, float_nan(dat_dict['mean']) ) )
pos += step_size
- return result
+ return { 'data': result }
class BigBedDataProvider( BBIDataProvider ):
def _get_dataset( self ):
@@ -654,7 +678,7 @@
float( feature[5] )]
rval.append(payload)
- return { 'data_type' : 'vcf', 'data': rval, 'message': message }
+ return { 'data': rval, 'message': message }
class GFFDataProvider( TracksDataProvider ):
"""
--- a/lib/galaxy/web/controllers/tracks.py Mon Sep 12 14:51:26 2011 -0400
+++ b/lib/galaxy/web/controllers/tracks.py Tue Sep 13 16:55:17 2011 -0400
@@ -530,15 +530,9 @@
data_provider = data_provider_class( converted_dataset=converted_dataset, original_dataset=dataset, dependencies=deps )
# Get and return data from data_provider.
- data = data_provider.get_data( chrom, low, high, int(start_val), int(max_vals), **kwargs )
- message = None
- if isinstance(data, dict) and 'message' in data:
- message = data['message']
- tracks_dataset_type = data.get( 'data_type', tracks_dataset_type )
- track_data = data['data']
- else:
- track_data = data
- return { 'dataset_type': tracks_dataset_type, 'extra_info': extra_info, 'data': track_data, 'message': message }
+ result = data_provider.get_data( chrom, low, high, int(start_val), int(max_vals), **kwargs )
+ result.update( { 'dataset_type': tracks_dataset_type, 'extra_info': extra_info } )
+ return result
@web.json
def save( self, trans, **kwargs ):
--- a/static/scripts/trackster.js Mon Sep 12 14:51:26 2011 -0400
+++ b/static/scripts/trackster.js Tue Sep 13 16:55:17 2011 -0400
@@ -443,9 +443,9 @@
$.extend(extra_params, {start_val: cur_data.data.length + 1});
}
else if (req_type === this.BROAD_DATA_REQ) {
- // Set query low to be past the last feature returned so that an area of extreme feature depth
- // is bypassed.
- query_low = cur_data.data[cur_data.data.length - 1][2] + 1;
+ // To get past an area of extreme feature depth, set query low to be after either
+ // (a) the maximum high or HACK/FIXME (b) the end of the last feature returned.
+ query_low = (cur_data.max_high ? cur_data.max_high : cur_data.data[cur_data.data.length - 1][2]) + 1;
}
//
@@ -463,6 +463,9 @@
// Update data and message.
if (result.data) {
result.data = cur_data.data.concat(result.data);
+ if (result.max_low) {
+ result.max_low = cur_data.max_low;
+ }
if (result.message) {
// HACK: replace number in message with current data length. Works but is ugly.
result.message = result.message.replace(/[0-9]+/, result.data.length);
Repository URL: https://bitbucket.org/galaxy/galaxy-central/
--
This is a commit notification from bitbucket.org. You are receiving
this because you have the service enabled, addressing the recipient of
this email.
More information about the galaxy-commits
mailing list