v1.2.0

• Fix for bismark not being a supported aligner.
• Run ataqv (https://github.com/ParkerLab/ataqv) to calculate additional ATAQ-seq quality control
  metrics.
• Workaround for some bcbioRNASeq plots failing with many samples when interesting_groups is not set.
• Add known_fusions parameter for passing in known fusions to arriba.
• Fix for tx2gene not working properly on some GTF files.
• Sort MACS2 output with UNIX sort to avoid memory issues.
• Run RiP on full peak file for ATAC-seq.
• Run ataqv on unfiltered BAM file with the full peak file.
• Run peddy on the population variant file, not the individual sample level file if joint calling was done.
• Add STAR to MultiQC metrics.
• Throw an error if STAR is run on a genome with alts.
• Don't run bcbioRNASeq if there is only one sample. Thanks to @kmendler for the suggestion.
• Improve arriba sensitivity by setting --peOverlapNbasesMin 10 and --alignSplicedMateMapLminOverLmate 0.5 when
  running STAR (see suhrig/arriba#41).
• Make TPM and counts files from tximport automatically.
• Use --keepDuplicates when making the Salmon index. This keeps transcripts that are identical in the index instead of
  randomly choosing one. This helps when comparing to other ways of quantifying the transcripts, ensuring all of
  the transcripts are represented.
• Remove unnecessary "quant" subdirectory for Salmon runs. This allows MultiQC to properly name the samples.
• Ensure STAR log file is propagated to the upload directory.
• Fix issue with memory not being specified properly when running bcbio_prepare_samples.py.
• Run tximport automatically and store TPM in project/date/tpm and counts in project/date/counts.
• Calculate ENCODE quality flags for ATAC-seq. See https://www.encodeproject.org/data-standards/terms/#library for a
  description of what the metrics mean.
• Fix for command line being too long while joint genotyping thousands of samples.
• Fix for command line being too long when running the CWL workflow with cromwell.