This repo contains a wrapper for the Scallop transcriptome assembly tool.
The wrapper is written in perl but requires an installation of the following applications:
If scallop, gffcompare and gtfcuff are not in your path they can be specified on the command line using the flags
--scallop_path, -gffcompare_path and --gtfcuff_path respectively.
The script ScallopAdvising.pl requires that the following options be given:
--working_dir-- a folder in which temporary files will be stored--input_bam-- the mapping of RNA-seq reads to the reference genome--output_gtf-- the output file--reference-- the reference transcriptome corresponding to the reference genome used
Then a list of configuration files should be supplied (as described below). An example of the full command is
./ScallopAdvising.pl --working_dir temp/ --input_bam in.bam --output_gtf out.gtf --reference ref.gtf default.config config1.config
optionally the --log_file parameter can be used to output a TSV file that lists the AUC for each of the input configurations.
The configuration files are flat files on which each line contains one tunable parameter and its corresponding value. Non-specified tunable parameters retain the default value.
An example of the default parameter configuration would be:
max_dp_table_size 10000
max_edit_distance 10
max_intron_contamination_coverage 2.0
max_num_exons 1000
min_bundle_gap 50
min_exon_length 20
min_flank_length 3
min_mapping_quality 1
min_num_hits_in_bundle 20
min_router_count 1
min_splice_boundary_hits 1
min_subregion_gap 3
min_subregion_length 15
min_subregion_overlap 1.5
min_transcript_length_base 150
min_transcript_length_increase 50
uniquely_mapped_only 0
use_second_alignment 0
Information about how to obtain the data used for validating parameter advising for transcript assembly can be found in the publication_results/ directory.
There you will find links and scripts to help retreive the necesary input to recreate the results found in the manuscript.
#Training
The information found in the training/ folder can be used to retrain either of the advisor set for a new set of data (though not neccesary).