eggd_starfusion v1.0.0

What does this app do?

Runs STAR-Fusion (https://github.com/STAR-Fusion/STAR-Fusion; https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1842-9), a tool which identifies candidate fusion transcripts from RNA alignments, from its official Docker image. The Docker image is provided by the user as an app input. This app is specifically set up to process the 'Chimeric.out.junction' files produced by the earlier STAR-Aligner alignment step. It produces files of candidate fusion regions.

What inputs are required for this app to run?

• The DNA Nexus file ID of a junction file, produced by STAR-Aligner
  • The file should be named in the format 'sample_name.chimeric.out.junction'
  • The file name is split on '.' and the first field used to name the output files
• The DNA Nexus file ID of a STAR genome resource, which should be a compressed '.tar.gz' file - from https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/
• The DNA Nexus file ID of a saved STAR-Fusion Docker image, which should be a compressed '.tar.gz'
• Additional parameters can be passed using 'opt_parameters', which should be a space-delimited string. If additional parameters aren't passed, the defaults will run.
  • If you're running 2 or more optional parameters, it will be formatted like so: -iopt_parameters="--opt_a data --opt_b more_data"
  • Options available are detailed at the bottom of the page under 'Appendix 1'.

How does this app work?

• Downloads and unzips/untars the STAR genome resource
• Downloads the sample junction file
• Loads and runs the STAR-Fusion Docker image using default settings
  • Production versions of this app will need to point to a controlled Docker image in 'references' on DNAnexus to ensure that the same version is run each time
• Prefixes output file names with the sample name
• Uploads the output files to DNA Nexus

What does this app output?

• The app generates:
  • starfusion_predictions: the predictions file prefixed with the sample name (extracted from the StarAligner junctions file), and ending '.fusion_predictions.tsv'
  • starfusion_abridged: the abridged version of the predictions file, prefixed with the sample name (as above), and ending '.fusion_predictions.abridged.tsv'

Appendix: Additional options

Options which may be useful

Further options available to change in STAR-Fusion, and obtained by running 'STAR-Fusion -h' inside an interactive Docker, are as below. For more information about the purpose of each parameter, see the 'help' messages for each parameter at https://github.com/STAR-Fusion/STAR-Fusion/blob/master/STAR-Fusion:

STAR program configurations:

• --STAR_max_mate_dist INT
• --STAR_SJDBoverhangMin INT
• --STAR_SortedByCoordinate
• --STAR_onepass

Stitching overlapping reads settings:

• --STAR_peOverlapNbasesMin INT
• --STAR_peOverlapMMp INT

STAR chim multi-map opts:

• --STAR_chimMultimapScoreRange INT
• --STAR_chimMultimapNmax INT
• --STAR_chimNonchimScoreDropMin INT
• --STAR_outSAMattrRGline STR

Chimeric read filtering parameters:

• --min_pct_MM_nonspecific|M INT

Fusion transcript filtering:

• --min_junction_reads INT
• --min_sum_frags INT
• --require_LDAS 0|1
• --max_promiscuity INT
• --min_pct_dom_promiscuity INT
• --aggregate_novel_junction_dist INT
• --min_novel_junction_support INT
• --min_spanning_frags_only INT
• --min_alt_pct_junction FLOAT
• --min_FFPM FLOAT
• --no_remove_dups
• --skip_EM
• --skip_FFPM

Turn off specific fusion filters:

• --no_annotation_filter
• --no_RT_artifact_filter
• --no_single_fusion_per_breakpoint

Downstream analysis of fusion candidates:

• --extract_fusion_reads

Miscellaneous options:

• --verbose_level INT
• --max_sensitivity
• --full_Monty

Options which are irrelevant or app-controlled

The below options only control where files are output and how resources are allocated, and are either not used, or are managed by the app:

• --output_dir
• --CPU
• --STAR_PATH
• --STAR_limitBAMsortRAM
• --tmpdir
• --STAR_use_shared_memory
• --STAR_LoadAndExit
• --STAR_Remove
• --outTmpDir STR

The below options are unlikely to be needed, because the workflow relies on a separate, StarAligner step being run to generate an initial predictions file, and a later FusionInspector step which examines coding effect after validating:

• --left_fq
• --right_fq
• --run_STAR_only
• --examine_coding_effect
• --no_filter
• --FusionInspector STR (inspect/validate)
• --denovo_reconstruct
• --misc_FI_opts STR
• --version
• --samples_file STR (intended for single-cell)

This app was made by East GLH