a bV@sddlmZddlZddlZddlmZddlmZmZm Z m Z m Z ddl m Z ddlmZe e jZiZGdddZd d Zd d Zd dZddZGdddZdS))with_statementN) defaultdict) fastaparser genes_parser reportingqconfigqutils) get_logger) run_parallelc@seZdZdddZdS)FeatureContainercCs||_||_g|_i|_dSN)kindfpaths region_listchr_names_dict)selfrrrD/home/psgendb/BIRCHDEV/pkg/quast-5.2.0/quast_libs/genome_analyzer.py__init__szFeatureContainer.__init__N)r )__name__ __module__ __qualname__rrrrrr sr cCstSr )ref_lengths_by_contigsrrrrget_ref_aligned_lengthssrcCsi}|D]&}|j|vr$|j||j<qd||j<qt|dkrt|dkr||djdur|}tjd||dj|fdd|D]}||_|||j<qnHtdd|Drtjd |ddnd|vrtjd |dd|S) z returns dictionary to translate chromosome name in list of features (genes or operons) to chromosome name in reference file. NrzReference name in file with genomic features of type "%s" (%s) does not match the name in the reference file (%s). QUAST will ignore this issue and count as if they match. indentcss|]}|duVqdSr r).0chr_namerrr 8z)chromosomes_names_dict..z{Reference names in file with genomic features of type "%s" do not match any chromosome. Check your genomic feature file(s).zSome of the reference names in file with genomic features of type "%s" does not match any chromosome. Check your genomic feature file(s).)seqnamelenpoploggernoticeallvalueswarning)featureregions chr_namesZregion_2_chr_nameregionr rrrchromosomes_names_dict"s4  *  r/c2 st|}t|}t} tt} tdt||t j ||d} t j rX| } n| d} t j | stjd| ddddSi} |D]\}}dg|d | |<qt|}tt|d d d d }g}g}|D] \}\}}||||qdgt|dgt|i}t|}t jr<|r`r"z%process_single_file..T)keyreversezAnalysis of genes and/or operons files (provided with -g and -O) requires extensive RAM usage, consider running QUAST without them if memory consumption is critical.| z?Something went wrong and chromosome names in your coords file (zS) differ from the names in the reference. Try to remove the file and restart QUAST.r#startendcontigstart_in_contig end_in_contigz _gaps.txtz /dev/nullw  gaps_count_full_partialZ_genomic_features_z.txtz%s %s %s %s %s )zID or #ZStartZEndTypeContigz3================================================== z# Fz%s %d %d complete %s operon,csg|]}|qSr)format_gene_info)rblock)r.rr r"z'process_single_file..cSs|jSr )r9)rJrrrr2r")r3z%s %d %d partial %s zAnalysis is finished.csg|] }|qSrrridx)features_in_contigsrrrKr"csg|] }|qSrrrL)operons_in_contigsrrrKr")8rlabel_from_fpathlabel_from_fpath_for_fnamedictrintr&info index_to_strospathjoinruse_all_alignmentsisfileerroritemsr read_fastasorted enumerateappendr$memory_efficientr*opensplitstrip AlignedBlockrangekeyssumspace_efficientendswithremove analyze_gapswrite min_gap_sizestrrFieldsGENESOPERONSrrloweridnumberr#r:r9rIminmaxmin_gene_overlapclose)2 contigs_fpathindexcoords_dirpathgenome_stats_dirpathreference_chromosomesns_by_chromosomes containersassembly_labelZcorr_assembly_labelresults ref_lengthsZcoords_base_fpath coords_fpathZgenome_mappingr chr_lenZ contig_tuplesZsorted_contig_tuplesZsorted_contigs_namesZ contigs_orderrMname_Zaligned_blocks_by_contig_nameZgene_searching_enabledZ coordfilelines1e1s2e2Z contig_nameirAZ gaps_fpathZ gaps_fileZ cur_gap_size container total_fullZ total_partialZ found_fpathZ found_fileZ found_listZ gene_blocks contig_idZcur_feature_is_foundZ cur_blockZ contig_infounsorted_features_in_contigsunsorted_operons_in_contigsr)rNrOr.rprocess_single_fileBs           4      &  J       $   $   rc3 stj|tjddlm}|jr0tjdt t dtj sXt t |\}tjd} t| d} g|D]\} } t| g| q|stjddd |rЈt|d ntjd dd D]} | jsq| jD]}| jt|| j7_qt| jdkrXtjd | jd dd | d| jdddqtdtt| jd| jd| d| jdtt| jdt| j| jt | _!qd\}}|D]v}t"#|}d}D]B} | jd krt| j}|$t"j%j&t| jn|t| j7}q|r|}|$t"j%j'|qi}i}i}i}g}g}g}tj(}t)t|tj*}fddt+|D}t,t-||dd\}|t|t7}|t_(st d| .dSD]&fddt/tDt0<q| dD]\\}} t1t0|}!| d|dt| dd t| t|d!t|!d"q| d| d#t|d$| d%ttj2d| d&ttj3d$| d$| d'd(| d'd)| d*t4||D]t\}\}"}#}$}%}&t56|}'|$||<|#||<|&||<|%||<|t7|$|t7|&|"d+}(|"t"j%j8d,})|"t"j%j8d-}*|"t"j%j9d,}+|"t"j%j9d-},t"#|}| d.|'dd/|:t"j%j;|:t"j%j<|(f|t=|:t"j%j;t"j%j8|)|*ft"j%j9|+|,ffD]V\}-}.}/|.durD|/durD| d0d1n&| d0|.|/f|$|-d2|.|/fq| dq| .tj>rdd3l?m@}0|r|0A||d4|||r|0A||d5||tjBrd6d7lCmD}1dd8lEmF}2|r2|1G|||d9d:|1H||||2|d;d:|1I||d<d=|r||1G|||d>d5|1H||||2|d?d5|1I||d@dA|1jI||dBdCdDdEt dFS)GNr)search_references_metarawzRunning Genome analyzer...zgenome_info.txtr>zbNo file with genomic features were provided. Use the --features option if you want to specify it. rrrGzPNo file with operons were provided. Use the -O option if you want to specify it.zNo genomic features of type "z" were loaded.zGenomic features of type "z " loaded: Noner?z Loaded z genomic features of type ""r0c s"g|]\}}||fqSrr)rr{rz)rr|r}rr~rrrKDszdo..T)filter_resultsz.Genome analyzer failed for all the assemblies.csg|]}|qSrr)rr)refrrrrKPr"zreference chromosomes:  z (total length: z bp, ztotal length without N's: z bp, maximal covered length: z bp) ztotal genome size: z zgap min size: zpartial gene/operon min size: z8%-25s| %-10s| %-12s| %-10s| %-10s| %-10s| %-10s| %-10s| )assemblygenomeZ duplicationZgapsgenespartialoperonsr)r fractionratiorur rr rzy======================================================================================================================== rArBrCz%-25s| %-10s| %-12s| %-10s|z %-10s| %-10s|)-rz %s + %s part) html_saverfeaturesrr)plotter)contigs_aligned_lengthsz/features_cumulative_plotzgenomic featuresz/features_frcurve_plotz/complete_features_histogramz# complete genomic featuresz/operons_cumulative_plotz/operons_frcurve_plotz/complete_operons_histogramz# complete operonsz/genome_fraction_histogramzGenome fraction, %d) top_valuezDone.)JrVrWrXraligner_output_dirname quast_libsris_quast_first_runr&print_timestamp main_infoisdirmkdirrget_genome_statsrbr\r`r r'rrrget_genes_from_filerr$r*rmrTror/listrgrrget add_fieldrp REF_OPERONS REF_GENES_num_nf_errorsrv max_threadsr_r rryrfrrwrnrxziprname_from_fpathrhrqrr get_field MAPPEDGENOMEDUPLICATION_RATIOfloat html_reportquast_libs.html_saverrsave_features_in_contigs draw_plotsr rquast_libs.ca_utils.miscrgenes_operons_plotfrc_plot histogram)3 ref_fpathaligned_contigs_fpathsoutput_dirpathZ features_dictZoperons_fpaths detailed_contigs_reports_dirpathr}r genome_sizeZ result_fpathZres_filer+ feature_fpathrfpathZ ref_genes_numZref_operons_numrzreportZgenomic_featuresZfiles_features_in_contigsZ"files_unsorted_features_in_contigsZfiles_operons_in_contigsZ!files_unsorted_operons_in_contigsZ genome_mappedZfull_found_genesZfull_found_operonsZ num_nf_errorsn_jobsparallel_run_argsZresults_genes_operons_tuplesr r aligned_lenrrrNrrO assembly_namerAZ genes_fullZ genes_partZ operons_fullZ operons_partfieldfullpartrrrr)rr|r}rrrr~rdos6       &&    $        &        rc@seZdZdddZddZdS)reNcCs(||_||_||_||_||_||_dSr r8)rr#r9r:r;r<r=rrrrs zAlignedBlock.__init__cCs|j|j}}|j|jkr@|j|j}||kr8||7}n||8}|j|jkrz|jt|j|j}||krr||}n||}|jdt|dt|S)N:r)r<r=r9r:rwr;ro)rr.r9r:Z region_shiftZ region_sizerrrrIs     zAlignedBlock.format_gene_info)NNNNNN)rrrrrIrrrrres re) __future__rloggingrV collectionsrrrrrrrquast_libs.logr quast_libs.qutilsr LOGGER_DEFAULT_NAMEr&rr rr/rrrerrrrs      06