U ÔL_Ðã@stdZddlmZmZddlmZddlmZddlm Z e e ƒZ Gdd„deƒZ dd „Z d d „Zd d „Zdd„ZdS)zz Collection of functions for metagenomic pipeline for taxonomic classification Created on June 3, 2014 @author: namphuon é)ÚJoinÚJobPool)Ú SeppProblem)Ú HMMAlignJob)Ú get_loggerc@s8eZdZdZdd„Zdd„Zdd„Zdd „Zd d „Zd S) Ú JoinBlastJobsz¥ After all blast search jobs have finished on markers, we need to figure out which fragment goes to which marker. This join takes care of that step. cCst |¡dS)N)rÚ__init__©Úself©r ú+lib/python3.8/site-packages/sepp/scratch.pyrszJoinBlastJobs.__init__cCs(||_| ¡D]}| |jd¡qdS)NZ blastsearch)Ú root_problemÚ iter_leavesZadd_jobÚjobs)r r Úpr r r Úsetup_with_root_problems z%JoinBlastJobs.setup_with_root_problemc Csd|jjkrdStdd„|jj ¡Dƒƒ}|j ¡D]„}| ¡}t|tƒsNt ‚|jdkrh|jj  g¡|_|  d¡}| ¡D]<}||\}}|dks¢|||dkrz||d|f||<qzq4g}|  ¡D]2\}} | ddkræ|  |¡qÆ| djj |¡qÆd|jjd<t dt|ƒ¡dS)z4 Figure out which fragment should go to which markerzfragments.distribution.doneNcSsg|] }|df‘qS))NNr )Ú.0Únamer r r Ú !sÿz;JoinBlastJobs.figureout_fragment_marker..Z hmmsearchéz.Fragments %s are not scored against any subset)r Z annotationsÚdictÚ fragmentsÚkeysrZ get_parentÚ isinstancerÚAssertionErrorZget_soft_sub_alignmentÚget_job_result_by_nameÚitemsÚappendZ seq_namesÚaddÚ_LOGZwarningÚstr) r Z max_evaluesÚfragment_chunk_problemZ align_problemZ search_resÚkeyZ best_valueZprev_align_problemZ notScoredÚvr r r Úfigureout_fragment_markers6  ÿ  ÿÿ      ÿz'JoinBlastJobs.figureout_fragment_markercCsÊ| ¡dd„|jjDƒ}|D]¦}t|tƒs0t‚t| ¡ƒ}|j  |¡}t |jƒD]p\}}|||_|j d}t|t ƒs|t‚|  d¡|_|j dj|_|j ¡s°|j |j¡nd|_tƒ |¡qRqdS)zº Distributes fragments to alignments subsets with best score, and runs align jobs on those. Also, creates new chunks of fragments for better parallelism. cSsg|]}|jD]}|‘qqSr )Úchildren)rrZalgr r r rWsÿz)JoinBlastJobs.perform..ZhmmalignZhmmbuildTN)Zfigureout_fragment_subsetr r%rrrÚlenZ get_childrenrZdivide_to_equal_chunksÚ enumeraterrrZhmmmodelZinfileZbase_alignmentZis_emptyZ write_to_pathZfake_runrZ enqueue_job)r Z alg_problemsZ alg_problemZchunksZfragment_chunksÚir!Zajr r r ÚperformJs( ÿ    ÿzJoinBlastJobs.performcCs d|jfS)Nz!join search jobs for all tips of )r r r r r Ú__str__pszJoinBlastJobs.__str__N) Ú__name__Ú __module__Ú __qualname__Ú__doc__rrr$r)r*r r r r rs -&rcCsdS)z:Blast the fragments against all marker genes+16S sequencesNr ©ZconfigÚinputÚoutputr r r Úblast_fragmentstsr2cCsdS)z€Fixes the direction of all the reads by searching the sequences against each HMM and finding out which one works the bestNr r/r r r Ú fix_directionxsr3cCsdS)zGReverse a sequence to be in the same direction as marker sequences Nr )Zsequencer r r Úreverse_sequence}sr4cCsdS)z Reads the results for blast Nr )r0r r r Úread_blast_results‚sr5N)r.Zsepp.schedulerrrZ sepp.problemrZ sepp.jobsrZsepprr+rrr2r3r4r5r r r r Ús   e