oicr-gsi / starmulti

RNASeq STAR Alignment

The STAR pipeline takes in 2 gzip-ped FASTQ files from RNAseq sources (paired sequencing), trims and then process them through a set of steps to map the reads to the supplied reference genome, annotate Read Groups and index the bam file.

The pipeline implemented as a SeqWare workflow.

Overview of the pipeline

The below flowchart summarizes each of the components' functionality

Compile

mvn clean install

Usage

After compilation, test, bundle and install the workflow using the techniques described in the SeqWare documentation.

Workflow Parameters

Input/Output:

    input_file_1    string  input file with the first mate reads. Presently only one file is allowed
    input_file_2    string  input file with the second mate reads. Presently only one file is allowed
    index_dir       string  directory with STAR indexes, workflow needs it to align reads of specific length

    output_prefix   string  a root directory for outputting the results
    output_dir      string  specifies a subdirectory for outputting files
    manual_output   string  When false, a random integer will be
                            inserted into the path of the final file
                            in order to ensure uniqueness. When true,
                            the output files will be moved to the
                            location of output_prefix/output_dir
                            [false]Determines if randomly named subdir will be created in output directory tree

    queue           string  SGE cluster queue

Read Group Information (Supported so that User could override these if needed):

    rg_platform_unit    PU in Read Group annotation
    rg_library          LB in Read Group annotation	 
    rg_platform         LB in Read Group annotation	 
    rg_sample_name      LB in Read Group annotation	 
    rg_organization     Organization that performed the experiment
    sequencer_run_name  LB in Read Group annotation
    barcode	            if available, the sequence of a barcode
    lane                sequencer lane
    ius_accession	    File IUS accession

IUS = Indivisible Sequencing Unit

    star             path     points to STAR binary in a directory where the bundled STAR resides
    picard_dir	 dir      Directory with operation-specific jar files for picard tools
    star_aln_threads integer  Threads (Cores) requested for STAR on SGE cluster
    star-aln-mem-mb  integer  Memory (in Mb) allocated to STAR SeqWare job
    uniqMapQ         integer  Score assigned to reads aligned to a unique location (Unique mappers)
    multimap_max     integer  This is to ensure we get all multi-mappers, 
                              may be customized to limit number of reads 
                              mapped to multiple locations in the final bam
    sa_sparsed       integer  this is a memory-optimization parameter, we use the same as PMH folks are using
    
    additionalStarParams      If needed, user may supply additional STAR parameters

By default, STAR assigns 255 as mapping score to such reads, this may break downstream analyses (especially with GATK) so the workflow overrides this default value assigning 60 instead

Decider Parameters

Arguments to the decider

Input/Output:

    manual-output   Specifies how the output subtree is created (random number subdirectories created or not)
    index-dir       This should be supplied each time
    template-type   It is recommended to use WT, MR and other RNAseq template type(s)
    output-dir      subdirectory for outputting the results, will be created automatically if does not exist
    output-prefix   where do we want the results
    queue           SGE queue, normally not set but for STAR we need a queue that would be able 
                    to reserve multi-core nodes for parallel processing
    manual-output   Provision files into directory subtree with randomly named subdirectory 
                    (when set to FALSE)
    verbose         Request more to output more information when running the Decider

STAR Parameters:

    star-aln-threads      integer  Threads (Cores) allocated to STAR job
    star-aln-mem-mb       integer  Memory in Mb allocated to STAR SeqWare job
    additionalStarParams  string   User may supply additional STAR parameters

    read1-adapter-trim    string  Adapter trimming 
                                  i.e. TruSeq Universal Adapter (AGATCGGAAGAGCGGTTCAGCAGGAATGCCGAGACCG)
    read2-adapter-trim    string  Adapter trimming
                                  i.e. TruSeq Universal Adapter (AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT)

Read Group Data:

    rg-platform-unit    PU in Read Group annotation
    rg-library          LB in Read Group annotation
    rg-platform         PL in Read Group annotation
    rg-sample-name      SM in Read Group annotation
    rg-organization     CM Organization that performed the experiment [OICR]

    outSAMattributes (may be also None or noQS but we do not need this set to any of these values)	 
    twopassMode      type of two-pass alignment [Basic]
    readFilesCommand Command for assisting with reading the input files [zcat]
    outSAMmultNmax   Defines how multimapped reads are handled
    outSAMmapqUnique STAR uses 255 mapq for uniquely mapped reads. 
                     This is not good for downstream analysis (GATK disregards reads with mapq 255) [60]
    outSAMtype       May be Unsorted, SortedByCoordinate or both (two files per alignment will be produced)

Output files

File basename is constructed using Meta-data information obtained via Decider and includes File SeqWare ID, Donor, sequencer run, library, barcode and lane information.

FILE_BASENAME.Aligned.sortedByCoord.out.report.bam

FILE_BASENAME.Aligned.sortedByCoord.out.report.bai

Alignment file and it's index with Read Group information added, sorted by coordinate

Support

For support, please file an issue on the Github project or send an email to [email protected].