New

• completely rewritten in Julia!
  • significantly faster under the hood
• added alternative mass-sample-from-VCF mode of operation

Breaking

• removed barcode specs (length/count), turns out they weren't necessary

Breaking Changes

• replaces --lr-type with --segments
  • accepts an integer
  • --segments 1 uses the entire barcode (10X, tellseq)
  • --segments 3 is a 3-segment combinatorial (stlfr)
  • --segments 4 is a 4-segment combinatorial (haplotagging)
• replaces --length with --lengths
  • now uses the format --lengths R1,R2
  • e.g. --lengths 132,150
• default output type for haplotagging (--segments 4) is standard:haplotagging

New

• adds standard suffixes
  • standard:haplotagging for @SEQID BX:Z:ACBD VX:i:N
  • standard:stlfr for @SEQID BX:Z:1_2_3 VX:i:N
  • just standard is the nucleotide barcode @SEQID BX:Z:ATCG VX:i:N

The singleton logic was inverted and for that I am truly sorry 🙏 -- FIXED

Fixed a little oopsie in the final output files.

New

• --circular option to simulate molecules from linear FASTA sequences as though they are circular

Breaking changes

• none

Non-breaking changes

• [internal] the barcode generation system is Class-based (cleaner code)

Fixes

• the writing process opens the output files once in write mode, instead of many times in append mode
  • should speed up the file writing
• main is cleaned up a bunch due to new class system
• more errors caught, errors are printed nicer

Breaking Changes

• None, except the output .gff file is now gzipped

Non-breaking Changes

• writer task is now a Process (used to be a Thread)
  • the wgsim simulation is still a thread-based process
• reverted to calling wgsim.core() instead of as a subprocess (another possible speedup)
• longmoleculerecipe logic now writes a temp fasta file (speeds things along)
• N reads from a molecule are drawn from exponential distribution instead of lognormal and then transformed out of log
  • the log-unlog transformations were slowing things down significantly
• Set a minimum of 2 reads from a molecule prior to possible downgrade to singleton
  • more reliable singleton ratios

Fixes

• simplified thread-watching and backpressure
• should have hopefully addressed some of the noticeable slowdown between v1 and v2

The Ship of Theseus edition

This is an almost complete (>98%) rewrite of the simulator such that it can be considered something entirely different from the source material and inspiration, XENIA.

Breaking Changes

• somehow, none
  • the file outputs are a little different, so that would be considered breaking if Mimick was part of a pipeline with specific output expectations

Non-breaking changes

• output is now one pair of FASTQ files, rather than a pair for each haplotype

New

• molecules sharing a barcode can now span contigs and haplotypes
  • process described here
• singular GFF output file for mutations
• outputs a molecule manifest file that lists all molecules that were simulated and some important details about them

Internal

• completely rewritten simulator
• significantly improved multithreading
• one thread always reserved for writing final output files

PRs closed @pdimens in #1

Full Changelog: 1.3...2.0

New

• --seed to optionally set a random seed
• --singletons to specify a proportion of singletons (i.e. barcodes with only one read pair)

Fixes

• Barcodes now properly write to file

Breaking changes

None