## 3.5.2 (2020-07-13)
- made numpy install version dependent on python version

## 3.5.1 (2020-07-10)
- fixed bug related to slow/interrupted file writing 

## 3.5.0 (2020-01-27)
- changed internal setup to allow annotation of large input files (through queue file writing)
- added command-line option "--chunk" to control the number of lines per chunk for multiprocessing (default 1000)
- added internal "annotate_single_peak" for easy integration into other tools

## 3.4.0 (2019-09-05)
- fixed improper handling of feature attributes with multiple tag:value pairs using identical tags. The values will now be written as comma-separated lists per tag.
- added +/- options for --strand (as addition to same/opposite)
- fixed creation of output folder to follow config input

## 3.3.4 (2019-07-01)
- removed the upper x-limit for "distance to feature" plots in the UROPA summary script (bug)
- made small changes to pie-charts as ggplot does not fill the circle 100%

## 3.3.3 (2019-06-06)
- fixed a bug in the UROPA summary script. The script was updated to reflect changes in query numbering. 

## 3.3.2 (2019-05-22)
- fixed confusion of feat_ovl_peak and peak_ovl_feat. The functionality is now as stated in the documentation, which is: feat_ovl_peak == 1.0 == PeakInsideFeature and peak_ovl_feat == 1.0 == FeatureInsidePeak. 
- rounding of small numbers will now print as many decimal places as needed to show fraction

## 3.3.1 (2019-05-21)
- fixed bug in subset_gtf not taking into account comments when subetting gtfs with header

## 3.3.0 (2019-04-28)
- added '--output-by-query' to allow for output .txt/.bed files per query (split from allhits results). This option can also be given in the config as "output_by_query".
- changed all underscores in input arguments to dashes to conform with standard CLI options. This change is backwards compatible, e.g. --feature_anchor can still be used for --feature-anchor'.

## 3.2.0 (2019-04-17)
- added functionality to view all available gtf-attributes in output files. Set by show_attributes = all.

## 3.1.4 (2019-04-13)
- fixed error due to insufficient handling of "chr"-prefix of both bed and gtf

## 3.1.3 (2019-04-09)
- fixed bug in calculation of 'relative location' occuring for peaks overlapping feature start/end but with anchor=center
- fixed py2 compatibility issues including:
  - changed import from "from uropa.utils import *" to "from .utils import *"
  - removed 'encoding' keys from read statements
  - forced float division

## 3.1.0 (2019-03-31)
- fixed missing sorting of gtf
- added handling of comment lines in gtf
- fixed insufficient backwards functionality for feature.anchor -> feature_anchor and other keys
- internal speed-up of priority queries by fetching hits from queries separately
- keys such as threads and prefix can now be given via config file
- re-added --summary

## 3.0.0 (2019-03-18)
- revision of functionality to fix several bugs and to simplify future problems
- new functionality: 
  - bed/gtf and single query annotation specification can be given via command line
  - show_attributes outside of query specification
  - additional name information per query
  - summary is always produced automatically
  - final and allhits are available in txt and bed format

## 2.0.4 (2019-01-30)
- fixed split command line call: should now work on different linux versions

## 2.0.3 (2019-01-10)
- fixed critical features bug, now empty or multiple features in config are possible


## 2.0.2 (2019-01-09)

- remove feature print
- finalhits additionally in bed format available

## 2.0.2-alpha (2018-01-16)

- added summary file for invalid annotations
- added proper magic line and file encodings for python source files

## 2.0.0-alpha (2017-11-08)

- update uropa to python 3 format with the aid of python 2to3

## 1.2.1 (2017-09-07)

- Rearranged package structure towards a full Pypi packages
- Renamed summary.R to utils/uropa_summary.R
- Renamed reformat_output.R to utils/uropa_reformat_output.R
- Reflected changes in documentation

## 1.2.0 (2017-09-05)

- Added an UpSetR plot to summary.R to cancel the need for Vennerable R package
- Added proper help to uropa2gtf.R script
- Reflected changes in documentations

## 1.1.2 (2017-09-04)

- Added proper help for axillary R scripts (exit code 0)
- Made call to reformat_output.R multi-threaded
- Cleaned some code, added citation

## 1.1.1 (2017-03-08)

- Bug fix for possible older versions of GNU sort, the order of results may have changed now
- Added protection against empty GTF files
- Clarified Tabix warning messages with proper genome coordinates.

## 1.1 (2017-01-30)

- Introduced parameter -p/--prefix, replaces -o/--output
- Replaced parameter --no-comment by --add-comments
- Renamed output files to prefix_allhits.txt, prefix_finalhits.txt and prefix_besthits.txt
- Updated summary script for new gridExtra requirements (rows=NULL instead of show.rownames=FALSE)
- Fixed bug in log file handling
- Fixed bug in internals features, now works for distances larger than default distance
- Removed summary config after usage, removed split_peaks subdirectory in case of multi threading

## 1.0 (2016-11-22)

- Fixed bug for log files in subdirectories
- Added new parameter --no-comments to skip comments in output tables
- Moved reduced annotation to output folder
- Minor changes in parameter descriptions

## 0.3 (2016-10-10)

- Changed function "valid_fsa" to check for strand[ignore,same,opposite]+filter.attributes + hit.feature//mk
- Added new parameters debug, log and threads.
- Added config file help as epilog.
- Parsing queries now accounts for empty ("") values.
- Invalid features result in a Warning, not in a program stop.
- Invalid queries (with more than one attribute.filter or more than two distances) are deleted, program continues on other queries.
- Annotation logic is moved out of main script
- Multiprocess splitting of peaks is save against peak loss now
- Multiprocessing falls back to one thread if split did not work
- Cleaned up some code lints, reformatted with autopep8

## 0.2 (2016-10-09)

- When Priority = True but no hit validates query No.1 , the queries No 2,3 if existent, will be read for finding a valid hit.
- The command line arguments -i, -o are flexible in position "
- The non overlapping peaks are saved in All_hits and in Best_hits table.
- All and Best hits will contain all peaks, with NAs in Best_tab when hit is in D > config "distance".
- Query column shown only if more than 1 query given.
- Check for all keys not to have EMPTY values.
- Check distance of p.center -gene.center,start,end  for valid_dist of hit.
- Fix cases of checking peak columns(make 3 'if's for diff sizes).Also read strand and replace to None if strand == ".".
- Extract attribute value one by one from all queries and give "not.found" if key doesn't exist.(get_hit_attribute)
- When correct_dir but best_hit or All_hits already written, just "continue", no NAs bcs it overwrites best hit of same peak,or keeps NA while there is also hit.
- Hits from all queries are visible in All_hits, per query, and in Best Hits the best per Query. BestBest hits table has the best of all queries.
- Features of gtf are extracted and replaced as default if no key given. If wrong feat_value given ->Error proposing gtf features.
- Fix NAs when pr=True and quer>1
- Add genomic_location, min_pos of the Dmin, overlap ratio, in the output tables.Remove internal_features key, but Find 'inside' or 'includefeature'(genom_location)
- Add 'filter.attribute' &  attribute.value' as extra keys in config for validating a hit.
- Add possibility of keeping different cut-off distance for Upstream and Downstream direction of peak when Distance = [x,y]
- Call script for Summary graphs and create file after creation of tables.

## 0.1

- Initial version
