CHANGES.txt

-- Version history --

5.3.0
      1. Compatibility changes:
       - Linux 32-bit systems are not supported anymore.
       - Python2 compatibility is not guaranteed anymore.

      2. New option:
       - "--no-krona" for not drawing Krona pie charts (MetaQUAST de novo mode).

      3. Update in embedded tools:
       - miminap2 updated from v.2.24 to 2.28 (alignments should remain the same);
       - new GeneMark license files.

      4. Minor fixes:
       - proper assembly coloring in the MetaQUAST HTML report for more than 14 assemblies;
       - catching the Krona charts crashes prematurely;
       - proper checking of user BED files in the read alignment mode;
       - cosmetic changes in warning/error messages.

      5. Documentation:
       - actualized copyrights, citations and support email.


5.2.0
      1. Core alignment changes (might affect all alignment-based metrics!):
       - new version of minimap2 (2.24; pre-installed minimap2 in PATH accepted if >=2.19);
       - default bandwidth for chaining and base alignment (minimap2's '-r' option) increased
         from 85 to 200 (now it is controlled via QUAST's "--local-mis-size" option unless
         "--large" is specified, in the latter case the default minimap2 value is used);
       - 'asm10' preset is used instead of 'asm5' for "--min-identity" in range [95.0, 99.0).

      2. New metrics:
       - auN, auNG, auNA, auNGA (areas under the Nx/NGx/NAx/NGx curves; for more detail see
         https://lh3.github.io/2020/04/08/a-new-metric-on-assembly-contiguity or the manual).

      3. New options:
       - "--local-mis-size" for setting minimal local misassembly size (default is 200, was 86);
       - "--report-all-metrics" for keeping the same content (list of metrics) in the main report
         independently of inputs/options.

      4. MetaQUAST change:
       - preserving explicitly specified reference genomes in the reports (they were previously
         hidden if nothing aligned to these references).

      5. Major fixes:
       - substantial speed up of the BSS algorithm (important for large genomes or MetaQUAST);
       - preserving assembly names in reports (hyphens were changed to underscores before);
       - "--no-gzip" option removed (it was never used in the pipeline); incorrectly set
         .gz extensions also removed (used_snps and gff files are not compressed by QUAST);
       - preserving explicitly specified reference genomes in metaQuast reports
         (were hidden if nothing aligned to these references).

      6. Minor fixes (rare crashes or slightly incorrect results in specific cases):
       - total reads number (was incorrectly calculated in case of secondary, supplementary,
         and duplicate reads);
       - proper processing of filepaths with commas;
       - check correctness of existing BED files and overwriting if corrupted.

      7. Updates in embedded tools:
       - updated Augustus (BUSCO dependency) to 3.3.3;
       - new GeneMark license files.

      8. Cosmetic changes in warning/notice/error messages.


5.1.0rc1 (remained as a release candidate)
      1. MetaQUAST changes:
       - new option: "--reuse-combined-alignments" for reusing alignments against the
         combined_reference in the subsequent runs_per_reference analysis stages;
       - new default: "--min-identity" default value set to 90% for both combined_reference
         and runs_per_reference stages. Compare with 95% default in regular QUAST;
       - improved no-ref mode: download the best (less fragmented, more complete) available
         assembly; search references with respect to strain and isolate; speed up downloading;
         fixed some internet connection issues; do not limit number of reference fragments
         when --ref-list is used.

      2. New option:
       - "--x-for-Nx" for reporting Nx, Lx, etc metrics for specific value of 'x' in
         addition to N50, L50, etc. The default value is 90. The previous non-changeable
         default was 75.

      3. New way of calculating old metrics:
       - Num mismatches/indels per 100 kbp (now computed with respect to the total number
         of aligned bases in the _assembly_ rather than in the _reference_ as before, may be
         important when Duplication ratio is way above 1);
       - Do not report a misassembly/break between the first and last alignment block of
         a contig if it covers more than 95% of a cyclic chromosome/plasmid (prokaryotes only).

      4. Critical fix:
       - Python 3.8+ support (cgi.escape replaced with html.escape)

      5. Small fixes (rare crashes or slightly incorrect results in specific cases):
       - GC calculation (zero division due to rate side effects);
       - duplication ratio (fractional overestimation due to Ns stretches in some contigs);
       - HTML report heatmap colors for partial BUSCO genes (red-blue color switch);
       - use of provided BAM files (--bam option not working properly);
       - postprocessing of minimap2 mediocre aligments (good pairs of alignments with a
         strecth of mismatches/indels/Ns in between were skipped due to low averageg IDY).

      6. Updates in embedded tools:
       - new version of SILVA database (138.1);
       - fixed links to BUSCO databases (v3/odb9);
       - new GeneMark license files.

      7. Cosmetic changes in warning/notice/error messages, pipeline steps order, etc.


5.0.2
      1. Fixed bug with missing genome features reference stats and plot in report.html

      2. Fixed bug related to newest versions of joblib (0.10 and higher).

      3. Fixed bug with some rare crashes of reads_analyzer module.

      4. Tiny fixes in error and warning messages.


5.0.1
      1. Using 'asm20' minimap2 preset for references with high divergence from the
         assembled organism (provided --min-identity is below 90%). As before, 'asm10' is
         used for --min-identity below 95% and 'asm5' for 95% and above.

      2. Fixed bug in using --split-scaffolds with MetaQUAST.

      3. Fixed bug in parsing genome sequences of GeneMark predicted genes.

      4. Fixed bug with crash of UpperBound creation when no paired-end reads are provided.

      5. Now FASTA entry names are considered as a sequence before the first space in the
         header line (">..."). Previously, the entire lines were considered.
         This change
           * shortens names of intermediate files (e.g. in k-mer-based metrics calculation)
           * simplifies using of standard annotation files (provided with --features/-g)

      6. Trying to use already installed minimap2, Glimmer, joblib, simplejson rather than
         distributions from the QUAST package (important for external QUAST installers).

      7. Improved documentation and error/warning/info messages.

      8. GeneMark licence files are updated.


5.0.0
      1. QUAST-LG mode is added ("--large") for evaluating large genomes!
         Significant speed up on large genomes achieved by the switch to fast Minimap2
         aligner and huge refactoring of the post processing bottlenecks in the QUAST code.
         The more adequate output is due to (1) improved handling of transposable elements
         (TEs) causing many false positive misassemblies in regular QUAST runs and (2) use
         of proper thresholds on minimal alignment, contig length, and extensive misassembly
         sizes.

      2. New module: upper bound assembly ("--upper-bound-assembly").
         We determine which part of the reference genome could be potentially reconstructed
         using a given set of reads. The algorithm takes into account zero covered regions
         and genomic repeats (identified with Red repeat finder). The constructed assembly
         is added to the evaluation to demonstrate the theoretical limits on the assembly
         completeness and contiguity quality metrics for the given genome and set of reads.

      3. New module: k-mer-based statistics ("--k-mer-stats").
         We identify unique k-mers in the reference genome (using KMC tool) and track
         their presence and relative location in assemblies. The percentage of the assembled
         k-mers is a novel completeness measure and the number of large inconsistencies
         (translocations or relocation with > 100 kbp difference in reference and assembly
         positions) is a novel correctness measure. By default, k is 101 bp and it can be
         specified with "--k-mer-size" option.

      4. Improved and extended gene prediction/annotation functionality:
       - Barrnap for rRNA genes prediction ("--rna-finding") is added;
       - BUSCO for finding conserved single-copy orthologs ("--conserved-genes-finding";
         Linux only) is added;
       - regular predicted genes (using GeneMark or Glimmer) are split into full and partial;
       - "--fungus" option is added for more accurate processing of fungus assemblies using
         GeneMark-ES and BUSCO;
       - "--features" option is added to replace "-G/--genes", it allows to count all genomic
         features from GFF or any specific feature type (e.g., 'CDS').

      5. Icarus updates:
       - changes in alignment viewers:
           * GC% track is added to the read coverage pane;
           * a button for highlighting all assembly misassemblies is added;
           * local misassemblies are now unchecked (hidden) by default.
       - static Circos plot of alignments ("--circos") is added;
       - chromosome names in the main menu are sorted in the human-friendly order now
         (e.g., chr1, chr2, ..., chr10 instead of chr1, chr10, chr2, ...).

      6. Improved reads support:
       - reads are now mapped to all assemblies and various alignment stats are reported;
       - single ("--single") and interlaced ("--12") reads are supported;
       - multiple read libraries are supported, including both paired-end ("--pe1/2/12")
         and mate-pair ("--mp1/2/12") libraries;
       - Oxford Nanopores ("--nanopore") and PacBio SMRT ("--pacbio") are supported;
       - ready SAM and BAM files can be provided both for reads mapped against assemblies
         ("--sam/bam") and reads mapped against the reference genome ("--ref-sam/bam");
       - reads stats can still be skipped by using "--no-read-stats" option.

      7. Modified processing of undefined nucleotides ('N'):
       - reference Ns are excluded from Genome Fraction computation (100% if all ACGT bases
         are covered);
       - assembly Ns are excluded from "Unaligned" and "partially unaligned length"
         computation;
       - scaffold gaps are now defined as simply a gap between alignments having at least 10
         consecutive Ns (affects "# scaffold gap size mis.", previously it was underestimated
         due to a strict threshold on the percentage of Ns in the gap sequence).

      8. MetaQUAST changes:
       - trying to download next best match if a reference genome is not found in NCBI
         (without references mode only);
       - link to the combined reference report is added to the main report HTML;
       - sample summary reports (TXT, TEX, etc) are renamed to exclude special characters in
         the filenames ('#', '%', etc).

      9. Changes and new metrics related to scaffold gap size misassemblies:
       - local scaffold gap misassemblies are added (local misassemblies caused by incorrect
         estimation of scaffold gap sizes);
       - contig and scaffold misassemblies are separated in the detailed misassemblies report
         (these scaffold misassemblies contain incorrectly estimated scaffold gap sizes
         exceeding scaffold-gap-max-size threshold or they are inversions/translocations caused
         by incorrect scaffolding).

      10. New and renamed options:
       - "--scaffolds" is renamed to "--split-scaffolds";
       - "--skip-unaligned-mis-contigs" is added to treat significantly unaligned (>50%) contigs
         with misassemblies as normal contigs (i.e. count their number of misassemblies in the
         misassembly-related metrics).

      11. Changes in the list of embedded third-party tools:
       - removed: GAGE, gnuplot;
       - replaced: MUMmer and E-MEM (new: Minimap2), Manta (new: GRIDSS);
       - added: BUSCO, Barrnap, KMC, Red.

      12. Fixed several minor bugs.


4.6.3
      1. Fixed crash of quast.py --test (introduced in v4.6.2).

      2. Fixed crash of BSS in MetaQUAST mode (introduced in v4.6.2).

      3. Proper float/integer division in both Python2 and Python3 (may affect
         the number of scaffold gap size misassemblies in Python2).


4.6.2
      1. Fixed relatively rare bug of BSS when using large --min-alignment.

      2. Improved check of previously generated results before reusing them.

      3. GeneMark licence files are updated.


4.6.1 
      1. Fixes in pip installation:
       - ignoring bdist_wheel and running regular installation instead;
       - no installation of aux files (LICENSE, Manual, etc);
       - proper error message about missing test_data dir (if "--test" is used).

      2. Fixed conda installation (no GeneMark installation if there is no home dir)


4.6.0 
      1. Switch from major.minor to major.minor.patch versioning.

      2. Python 3.6 is now supported (and all future Python 3.* versions).

      3. Best set selection algorithm is significantly sped up.

      4. GeneMark licence files are updated; expired licence error message is better
         handled now.

      5. Support email address is updated.

      6. Fixed several minor but nasty bugs.


4.5   1. Icarus updates:
       - always visible and scrollable Contig Info panel;
       - new color (dark green) for contigs without misassemblies but having at
         least 50% of their length in unaligned fragments.

      2. MetaQUAST updates:
       - parallel reference processing;
       - expandable "i/s translocations" and "possible mis. contigs" metrics on the
         main HTML report;
       - GC distribution plots for the combined reference;
       - new option ("--use-input-ref-order") to use originally specified order of
         the references in the summary plots (PDF/PNG/etc only);
       - stop processing with error message if any of the provided references is
         empty or has incorrect sequences (non-ACGTN characters or header-only).

      3. New plots are added (in static PDF/PNG/etc and interactive HTML formats):
       - Feature-Response Curve (FRCurve) for # misassemblies (both formats)
         and # genes/operons (static PDF/PNG/etc only);
       - Distribution of # contigs with GC % in a certain range (both formats);
       - MUMmer plot of assembly against reference mapping (static HTML only).

      4. Contig coordinates of the provided genes/operons annotations are saved to
         <output_dir>/genome_stats/<assembly_name>_genes.txt and _operons.txt.

      5. Broken scaffold assemblies (see "--scaffolds" option) are now filtered based on
         "--min-contig" value (default is 500 bp) similarly to regular assemblies processing.
         "Total length >= N bp" and "# contigs >= N bp" metrics are no more reported for
         these assemblies if N is less than the min contig threshold.

      6. In order to avoid confusion, "--meta" option was renamed to "--mgm" (MetaGeneMark).

      7. Internal overlaps between adjacent aligned blocks of a misassembled contig are
         always excluded. Previously the processing depended on "--ambiguity-usage" value.

      8. Small change in the installation strategy: if quast_libs directory has no write
         permission (e.g. after sudo setup.py install), external packages (e.g. SILVA 16S
         database; Manta SV caller) are downloaded to ~/.quast/ on the first use.

      9. Gnuplot version 5.0 is added to the installation package.

      10. GeneMark license files are updated.

      11. Fixed several minor bugs.


4.4   1. Icarus updates:
       - buttons for expanding overlapped alignments are added to detailed assembly tracks
         on Contig Alignment Viewers;
       - additional colors (available in all viewers) for:
           * ambiguous contigs,
           * alternative blocks of misassembled contigs,
           * misassembled contigs with >50% unaligned bases (see item #7 below).
       - better performance on low resolution screens.

      2. Python3 is now supported (versions 3.3, 3.4 and 3.5).

      3. Several new options are added:
       - "--scaffold-gap-max-size" for controlling maximum allowed size of scaffold gap
         inconsistency in the corresponding type of misassembly (--scaffolds only);
       - "--fragmented-max-indent" for controlling maximum allowed indent on each side
         of the reference fragments to consider translocation as fake one (see manual for
         details);
       - "--blast-db" for specifying custom BLAST database instead of SILVA 16S rRNA for
         searching probable references in MetaQUAST "without references" mode;
       - "--space-efficient" for removing or even not creating space consuming auxiliary
         files; all reports and plots (except Icarus viewers) are generated as usual;
       - "--significant-part-size" renamed to "--unaligned-part-size", see details below
         (see item #6 below).

      4. New metric, report and plot are added (MetaQUAST only, combined reference only):
       - # possible misassemblies (number of putative interspecies translocations
         in possibly misassembled contigs if each large unaligned fragment is supposed
         to be a fragment of unknown reference);
       - report of interspecies translocations number per each reference, saved to
         interspecies_translocations_by_refs_<assembly_name>.info under
         <output_dir>/combined_reference/contigs_reports/;
       - plot of all found and supposed (possible) interspecies translocations, saved to
         intergenomic_misassemblies.pdf under the same folder as above report.

      5. New detailed report is added:
       - all fully and partially unaligned contigs, their length, and unaligned parts are
         listed in <output_dir>/contigs_reports/contigs_report_<assembly_name>.unaligned.info

      6. Partially unaligned contigs are redefined in simpler and more logical way. Now
         all contigs having at least one unaligned fragment of length greater or equal to
         unaligned_part_threshold are considered as partially unaligned. The threshold is
         controlled by --unaligned-part-size option (default value is 500 bp).

      7. New processing of misassembled contigs which are mostly unaligned (> 50%):
       - "# half-unaligned with misassembly" metric is renamed to "# unaligned mis. contigs"
         and moved from detailed unaligned_report to detailed misassemblies_report
         (stored under <output_dir>/contigs_reports/);
       - "# unaligned mis. contigs" metric is added to all main reports (TXT, HTML, PDF).

      8. BED format is accepted for genes/operons annotations.

      9. Minor changes in HTML reports:
       - heatmap coloring is removed for metrics where best/worst values are undefined
         (GC %, # similar blocks, # contigs, etc);
       - split versions of scaffold assemblies are plotted with dashed lines (--scaffolds
         option only).

      10. Fixed issue of crashing E-MEM on OS X without gcc installed. Also, if E-MEM is
          already installed on the system, it will be used instead of embedded E-MEM.

      11. User-provided SAM/BAM files are now checked for correct chromosome names and
          automatically corrected if needed and correction is possible.

      12. Fixed incorrect behaviour of MetaQUAST on compressed assemblies in "without
          references" mode.

      13. Best set selection is improved (misassembly detection algorithm in case of multiple
          ambiguous mappings of contig fragments). Now BSS takes internal overlaps into
          account, so alignments with a short gap (in contig) are preferred over alignments
          having large internal overlap if all other things are equal.

      14. Fixed several minor bugs.

      15. Icarus citation is updated.

      16. Useful tips on running QUAST on large genome assemblies are added to FAQ section
          of the manual (see Q14).


4.3   1. Icarus updates:
       - predicted genes (see --gene-finding option) are now displayed in contigs (on both
         Contig Size and Contig Alignment Viewers);
       - zoom buttons added to read coverage panels insteads of log/normal scale switch;
       - small fixes for better performance on low resolution screens.

      2. setup.py is added for simplifying QUAST installation process.

      3. Heavy output files (lists of SNPs, GFF with predicted genes) are now compressed
         with gzip. This default behaviour may be cancelled by using --no-gzip option.

      4. Automatic suggestion to use --scaffolds option if assembly contains long continuous
         fragments of N's.

      5. Embedded samtools is replaced with sambamba which is significantly faster.

      6. Embedded bowtie2 is replaced with BWA-MEM which is more accurate.

      7. E-MEM is also working under OS X now (in addition to Linux).

      8. Removed limitation of not moving QUAST installation dir after the first use.

      9. Fixed bug causing creation of error.log file in current working directory.

      10. Fixed improper Duplication ratio calculation in some MetaQUAST runs.

      11. Requests to NCBI are switched to HTTPS which will be mandatory after September 30, 2016.

      12. SILVA database release is updated from 119 to 123 (it is downloaded on the first use).

      13. Fixed several minor bugs.


4.2   1. Icarus update -- improved read coverage panel: 
       - physical coverage is added (the coverage of the reference by the paired-end fragments,
         counting the reads and the gap between the paired-end reads as covered);
       - normal/log scale switcher is added;
       - samtools max coverage depth is increased, so now even very large coverages are shown.

      2. Icarus update -- search window is added:
       - searching contigs and genes/operons;
       - tooltip shows possible suggestions as one types.

      3. Icarus update -- Contig Alignment Viewer changes:
       - coordinates are counted for each chromosome separately;
       - all good alignments of ambiguous contigs are shown if --ambiguity-usage=all.

      4. Read coverage histograms are added for assemblies with SPAdes/Velvet-like contig
         naming style (i.e. all names are ..._length_X_cov_Y_...).

      5. Several new options are added:
       - "--references-list" for specifying list with reference names to be searched 
         and downloaded from NCBI (MetaQUAST only);
       - "--min-identity" for setting threshold on minimal IDY% of alignments (Nucmer 
         parameter, default is 95%);
       - "--ambiguity-score" for choosing how close good ambiguous alignments should be to
         the best one (default is 0.99);
       - "--sam" and "--bam" for specifying SAM and BAM files instead of/in addition to 
         raw reads. 

      6. Several new metrics are added:
       - Total aligned length (total size of contig fragments aligned to reference);
       - Icarus similarity statistics (in HTML report only, at least 2 assemblies are
         needed).

      7. MetaQUAST updates:
       - requests to NCBI database in no-reference mode are now repeated if error occurs;
       - Krona updated to v2.6, total aligned length is used for building Krona charts 
         (instead of Total length in previous versions);
       - possibly interspecies translocations are not reported if unaligned parts consist 
         mostly of Ns;
       - Total length and Largest contigs are substituted with their "aligned" version in
         summary HTML report (i.e. calculated for contig aligned fragments rather than 
         original contigs).

      8. HTML report is updated:
       - Genome statistics are moved to the top, Statistics without reference are moved to
         the bottom; 
       - reference line is added to GC content plot.

      9. GeneMark now outputs nucleotide and protein sequences for all found genes.

      10. Best set selection is improved (misassembly detection algorithm in case of
          multiple ambiguous mappings of contig fragments). Now BSS can find not only
          the very best set but all sets with scores close to the best one (controlled by
          --ambiguity-usage and --ambiguity-score options). Also several minor bugfixes
          are made and BSS become deterministic in case of equivalent scores.

      11. Embedded E-MEM aligner is updated to version 2.

      12. Significant refactoring of the source code. Most changes are in Icarus, 
          Contigs Analyzer, QUAST and MetaQUAST running scripts.

      13. Fixed several minor bugs.

      14. Icarus citation is updated.


4.1   1. Icarus update -- new visualization of contigs in both viewers:
       - all blocks are drawn with transparency, so overlaps are more visible;
       - controls for switching between overlapped blocks are added to Contig info panel;
       - y-coordinates offsets and shades of colors for neighbouring contigs are removed.

      2. Icarus update -- integration of Contig Size Viewer with Contig Alignment Viewer
         (if reference genome is available and thus alignment information is present):
       - contigs are colored according to their mapping on the reference (correct,
         misassembled, unaligned);
       - positions of the breakpoints in the misassembled contigs are shown;
       - full scale Contig info is now available with links to Contig Alignment Viewer;

      3. Nucmer aligner is replaced with significantly faster E-MEM (for Linux only).

      4. Significant speed up of misassembly detection algorithm in case of complicated
         genomes (with many short repeats).

      5. New option is added:
       - "--no-sv" for skipping structural variants calling and processing. Make sense
         only if reads are specified: in this case they will be used for building Icarus
         read coverage histograms only.

      6. Embedded Manta is updated to version 0.29.6.

      7. Reference line on cumulative plot is redesigned to be more consistent with
         assemblies lines (starts at 0, depends on number of chromosomes/plasmids).

      8. Fixed several minor bugs.


4.0   1. Icarus is added! Icarus stands for Interactive Contig Assessment browseR.
         Icarus complements QUAST output with interactive visualisations of assembly
         alignments to the reference genome, and all assembly features detected by 
         QUAST (e.g. misassemblies, genes). Icarus generates two types of viewers:
       - Contig alignment viewer (available if a reference genome is provided). Shows
         contig alignments to the reference, misassemblies, similarities between assemblies,
         genome annotations (if genes/operons are provided), read coverage along the genome
         (if reads are provided).
       - Contig size viewer. Draws contigs ordered from largest to smallest. Allows hiding
         of short contigs. Explicitly shows contigs of length N50, N75 (and NG50, NG75).

      2. Improved misassembly detection algorithm in case of multiple ambiguous mappings of
         contig fragments. The algorithm identifies best set of non-overlapping contig fragments
         mappings. The set minimises possible number of misassemblies, i.e. this algorithm 
         reduces number of false negatives (erroneously detected misassemblies).

      3. Several new options are added:
       - "--significant-part-size" for setting threshold on detecting partially unaligned 
         contigs with both significant aligned and unaligned parts (default: 500 bp as earlier).
       - "--fragmented" for notifying QUAST about fragmented reference genome (e.g. scaffold
         reference). QUAST will try to detect misassemblies caused by the fragmentation and
         mark them "fake".
       - "--unique-mapping" for forcing -a=one in QUAST run on the combined reference (MetaQUAST only).
       - "--sv-bedpe" for specifying BEDPE file with structural variations. Note that QUAST may
         create such BEDPE automatically using Manta SV caller if reads are specified.
         However, it is rather slow because include reads alignment to the reference.

      4. Fixed bug with skipping "broken" scaffolds in per reference runs of MetaQUAST.

      5. Slightly rearranged MetaQUAST summary HTML report. New order is: statistics,
         plots, table with links to per reference runs (the table was in beginning previously). 

      6. Fixed several minor bugs.
 
      7. Embedded samtools updated to version 1.3.

      8. Citation for MetaQUAST paper is updated, citation for Icarus paper is added.


3.2   1. The tool now accepts raw reads for improving quality assessment (experimental
      feature, try with care):
       - reads should be provided with --reads1 (or -1), --reads2 (or -2) options,
       - reads are aligned to reference genome using bowtie2 (embedded),
       - Manta structural variation (SV) calling tool (embedded) is run on bowtie2 output,
       - found SVs are used for classifying QUAST misassemblies into true ones and fake
         ones (caused by structural differences between reference sequence and
         sequenced organism). Fake misassemblies are excluded from "# misassemblies" metric
         and reported in novel "# structural variants" metric.

      2. HTML reports content is reformatted, especially in MetaQUAST reports:
       - GC % and all metrics based on genome length (NG50, NGA50, LGA75, etc) are excluded
         from the combined reference statistics (they don't make sense there);
       - N50 is replaced with more fair and comparable metrics such as "Total length >= 1000 bp",
         "Total length >= 10000 bp" in the main MetaQUAST report. Extended version still has N50.
       - N50 is also hidden in single-genome QUAST reports and exposed NGA50 instead.

      3. Scaffold gap size misassemblies are introduced. They are reported only when --scaffolds
         is used.

      4. Several new options are added:
       - "--memory-efficient" for running QUAST with minimal memory consumption (but significantly slower)
       - "--test-sv" for testing structural variants mode (see 1.)
       - "--silent" for minimal output to stdout (full verbose output is saved in the logs anyway)

      5. MetaQUAST output directory content is reformatted. Per reference reports are saved
         inside the directory <output_dir>/runs_per_reference, summary reports are under summary directory.
         The only top-level files are metaquast.log and report.html (summary HTML report with links to
         all subreports).

      6. Colors in HTML reports and PDF/PNG plots are synchronized, now they are the same
         for the same assemblies.

      7. Additional check for Matplotlib v1.1 (needed for drawing PDF/PNG plots since QUAST v3.1)

      8. Fixed several minor and major bugs.

      9. Citation for MetaQUAST paper added.


3.1   1. MetaQUAST:
       - more specific algorithm for reference searching and downloading, particularly
         only Bacteria and Archaea are downloaded;
       - Krona charts are added for showing taxonomic profile based on found references;
       - metric-level and misassemblies plots are added to summary HTML report;
       - better structure for text reports and plots in summary folder.

      2. Significantly reduced size of the installation package. This is done by removing
         BLAST binaries which are needed only for MetaQUAST run without references. On the
         first such run, BLAST binary for target OS will be automatically downloaded.

      3. Heatmaps are added to HTML reports.

      4. Separate install.sh and more complex install_full.sh scripts for installing regular
         versions of QUAST and MetaQUAST or extended one (with ability to run MetaQUAST without
         reference genomes).

      5. New option "--plots-format" for selecting output format for plots (PDF by default).

      6. Default number of threads is changed from 100% of CPUs to 25% of CPUs.

      7. Changes in one-letter options, including replacing confusing -T to -t for --threads
         and -M to -m for --min-contig.

      8. Skipping broken version of scaffolds from analysis if they are equal to original
         ones (see --scaffold option for details).

      9. Fixed several minor bugs.


3.0   1. Significant changes in MetaQUAST functionality:
       - if no references are provided, MetaQUAST downloads references from NCBI database
         based on best hits of assemblies alignments vs SILVA 16S rRNA database 
         (included in QUAST package);
       - multiple summarising reports are added: plots and text tables for each metric 
         (all assemblies vs all references in one file), histograms of # misassemblies 
         per reference for all assemblies separately, summary HTML-report with all metrics 
         per each assembly and each reference (expandable lines);
       - new metrics: interspecies translocations and # possibly misassembled contigs;
       - fixed handling of reference with more than one entry in FASTA file 
         (chromosome plus plasmids or multiple chromosomes);
       - option --reference now accepts directory (takes all references from this dir);
       - fixed handling of closely related species by using --ambiguity-usage 'all' on 
         combined reference run.
       
      2. Speed ups:
       - 5x-100x speed up in case of running QUAST without reference;
       - processing of large (> 50 Mbp) multi-chromosome references in parallel (per chromosome);
       - new speed up options: --no-check, --no-gc, --no-snps, --fast (a combination of other).

      3. More reports about misassemblies (in <output_dir>/contigs_reports/):
       - misassemblies_plot.pdf with histogram of misassembly types distribution per assembly.
       - contigs_report_<assembly_name>.mis_contig.info with brief details about misassembled 
         contigs only. 

      4. Improved and updated misassemblies detection algorithm:
       - more accurate algorithm for processing multiple ambiguous alignments;
       - using --ambiguity-usage value for processing internal overlaps between adjacent aligned
         blocks of a misassembled contig;
       - marking of local misassemblies with small inconsistency (<= 85 bp) as fake misassemblies
         (indels or mismatches) if gap on the contig is filled mostly with Ns;
       - option --extensive-mis-size/-x to set extensive misassembly size, i.e. min inconsistency
         size for relocations (default is 1000 as earlier);
       - option --min-alignment/-i to set minimum alignment length (Nucmer's parameter), default
         is 0.

      5. Fixed HTML-reports issues:
       - Y-axis coordinates on plots interactive hangover tooltips;
       - NAx plot small bug.

      6. GeneMark-ES is used for predicting genes in eukaryotic genomes instead of Glimmer-HMM.
      For using Glimmer-HMM, a new option --glimmer added.

      7. GeneMarkS incorporation is fixed. Previously it was run on predefined heuristic models based
      on GC content. Now it uses self-training module for getting the correct model.
      
      8. Fixed several minor bugs.

      9. GeneMark licenses is updated.

      10. Updated LICENSE (third-party tools details) and Manual.


2.3   1. Changed logic in misassembly computation. Fixed several minor bugs in misassembly
      detection algorithm and one major bug caused by linear representation of circular
      references and contigs in fasta format.

      2. Added contig alignment plots. See details in manual and in the QUAST paper (Fig. 1)

      3. Genome analyzer module (computation of genome fraction, duplication ratio,
      number of genes and operons) is parallelized.

      4. Option --test become an installation util analogue. It compiles all required binaries and
      checks correctness of QUAST and MetaQUAST execution on test datasets.

      5. Former plots.pdf is upgraded with report tables and renamed to report.pdf. Now it is
      a file with all tables and plots generated by QUAST.

      6. New option "--no-plots" for speeding up computation if plots are not needed.
    
      7. GeneMark license is updated, instructions for manual updating are added.

      8. Generation of misleading single-columns histograms is removed (when only single assembly
      file was specified).

      9. More error and exception handlers are added.

      10. Fixed bug with indel counting (caused slightly overestimated indels rate in some cases).
      
      11. Fixed several minor bugs.
       
      12. Code is refactored.


2.2   1. The tool now supports metagenomic assemblies. It accepts multiple references
      and produces several reports:
       - for all contigs and all input genomes merged into one,
       - separate reports for only contigs aligned to a particular genome,
       - for the contigs not aligned to any reference provided.

      Usage:
         metaquast.py contigs_1 contigs_2 ... -R reference_1,reference_2,reference_3,...

      All other options for metaquast.py are the same as for quast.py.

      2. MetaGeneMark is used to find genes in metagenomic assemblies.
      In metaquast.py by default, in quast.py with --meta option.

      3. In place of --allow-ambiguity, a new option --ambiguity-usage (-a) is introduced.
      The new option lets specify a way to process ambiguous regions:
      -a one, -a all or -a none.

      4. A new option --labels (or -l) allows to provide human-readable assembly names.
      Those names will be used in reports, plots and logs, instead of file names.
      For example:
         -l SPAdes,IDBA-UD

      if your labels include spaces, use quotes:
         -l "SPAdes 2.5, SPAdes 2.4, IDBA-UD"

         -l SPAdes,"Assembly 2",Assembly3

      5. Minor improvements of HTML reports.

      6. Fixed bugs in misassemblies detection algorithm.


2.1   Option --strict-NA is added to control computation of NAx/NGAx metrics.
      This option forces QUAST to break contigs by any misassembly event, 
      including local misassemblies (like in v.2.0). By default, QUAST v.2.1
      breaks contigs only by extensive misassemblies to compute NAx/NGAx 
      (like in v.1.*).

      Improvement of indels computation. QUAST now counts consecutive single nucleotide
      indels as one indel. Total length of all indels is also reported (equal to
      # indels metric evaluated with previous versions). Short (<= 5 bp) and long (>5 bp)
      indels are reported.

      Option --est-ref-size is added to set estimated reference size for computing NGx
      metrics in case a reference genome is not available.

      GAGE mode is parallelized.

      Fixed bugs in misassemblies detection algorithm.

      Fixed bugs in SNPs detection algorithm.

      Fixed bugs in processing circular chromosomes (affects Genome fraction, # genes,
      # operons).

      Fixed several minor bugs.


2.0   Significantly improved assessment of large genomes. Current limit on size of a
      reference genome is 536 Mbp PER CHROMOSOME instead of 536 Mbp TOTAL in the
      previous versions. Alignment to different chromosomes is performed in parallel.

      Changes in algorithm for evaluating Genome fraction, # genes and operons.
      Filtration of short, ambiguous, and redundant alignments is performed before
      the evaluation. Option --use-all-alignments is added for compatibility
      with 1.* versions.

      New algorithm for finding SNPs and indels.

      Ability to change colors, line styles, etc. in plots and content, metric names
      in reports.

      GlimmerHMM for predicting genes in eukaryotes.

      Gene Finding is parallelized and its run is controlled by --gene-finding option.

      Improvement of HTML-reports and plotting units.

      Fixed several bugs.


1.3   QUAST is now a multi-threaded tool: the most time-consuming step (alignment to
      a reference genome) is computed in parallel.

      A MacOS version of GeneMark.

      Significantly improved HTML-reports.

      More informative error messages.

      A simple logic for evaluating scaffolds.

      New metrics: duplication ration and largest alignment.

      More careful counting of misassemblies.

      Min contig threshold changed from 200 to 500.

      Fixed several bugs.


1.2   Indels and N's counting.

      More detailed statistics on misassemblies (classification in inversions,
      relocations, translocations, local misassemblies).

      More detailed statistics on partially unaligned contigs.

      Text reports are now available in LaTeX format.

      Python 2.5 is now supported.

      Fixed bug in reading genes annotations in GFF and NCBI formats.

      QUAST can be rerun on existing Nucmer alignments files.


1.1   Mismatches counting.

      Fixed bug in misassemblies counting (some inversions were omitted).

      GC content plot is logarithmically scaled.

      ORFs are not counted, GeneMark is added instead (for gene finding, only on Linux).

      Nucmer aligner parameters are changed (from IDY% = 80 to 95,
      i.e. all alignments become more robust).


1.0   Initial open source release!