making some well-founded assumptions about the nature of how
repeat regions in prokaryotic genomes appear in genome assemblies. While Unicycler can assemble a variety of different types of
sequencing reads (see long read-only and hybrid assembly sections
below), it does include a mode to run this optimized version of
SPAdes by itself. Thus, if the goal is production of de novo prokaryotic genomes that are as complete as possible from only Illumina short-read data, running Unicycler is advised.
3.6.2 Long Read-Only De
Novo Assembly
Algorithms for assembly using only ONT reads must deal with two
specific challenges inherent to Nanopore sequencing data: the substantially lower per-base Q score from Nanopore reads and the
abundance of erroneous indels, often errors in reporting of homopolymers (e.g., reporting “AAAAT” versus a true “AAAT”). ONT
assemblers can manage these issues by (a) easing the parameters by
which the reads are “stitched” together, allowing connections
between reads that would not otherwise be made with short reads
to create a consensus “rough” assembly and (b) correcting this
rough assembly by mapping the long reads back onto the consensus
assembly. Tools for assembling long-read sets include Miniasm
[49], Canu [50], Flye [51], and Unicycler Long Read-Only mode
(which reimplements Miniasm with some optimizations [48]); of
these, Flye is generally the most recommended long read-only
assembler [52]. Tools for polishing the resulting long read-only
assembly include Racon [53], Medaka [54], and Nanopolish [55];
Medaka is the latest entry into this rapidly growing and changing
field of algorithms and claims substantial accuracy and run time
improvements over Nanopolish.
3.6.3 Hybrid De Novo
Assembly
The primary disadvantage of using a singular sequencing technology is the necessary choice between either a highly accurate but
fragmented short read-only assembly or a highly complete but
relatively inaccurate long read-only assembly. When these technologies are used in conjunction with each other, the advantages of
each can be leveraged and their disadvantages mitigated, resulting
in a highly accurate and complete genome assembly. Such assemblies are called hybrid assemblies, since they combine more than
one type of sequencing technology. To date, Unicycler is generally
regarded as the best hybrid assembler for prokaryotic organisms
due to its optimizations at both the short-read assembly and longread assembly levels. Briefly, Unicycler begins with an optimized
SPAdes assembly and then proceeds to build “bridges” between
disconnected contigs using long-read data, both from a long-read
assembly and by mapping long reads directly onto the short-read
assembly. In a head-to-head benchmark comparison of Unicycler
versus other hybrid assemblers and assemblies using a variety of
sequencing technology, Unicycler was the clear top
performer [56].
416
Marco A. Riojas et al.
repeat regions in prokaryotic genomes appear in genome assemblies. While Unicycler can assemble a variety of different types of
sequencing reads (see long read-only and hybrid assembly sections
below), it does include a mode to run this optimized version of
SPAdes by itself. Thus, if the goal is production of de novo prokaryotic genomes that are as complete as possible from only Illumina short-read data, running Unicycler is advised.
3.6.2 Long Read-Only De
Novo Assembly
Algorithms for assembly using only ONT reads must deal with two
specific challenges inherent to Nanopore sequencing data: the substantially lower per-base Q score from Nanopore reads and the
abundance of erroneous indels, often errors in reporting of homopolymers (e.g., reporting “AAAAT” versus a true “AAAT”). ONT
assemblers can manage these issues by (a) easing the parameters by
which the reads are “stitched” together, allowing connections
between reads that would not otherwise be made with short reads
to create a consensus “rough” assembly and (b) correcting this
rough assembly by mapping the long reads back onto the consensus
assembly. Tools for assembling long-read sets include Miniasm
[49], Canu [50], Flye [51], and Unicycler Long Read-Only mode
(which reimplements Miniasm with some optimizations [48]); of
these, Flye is generally the most recommended long read-only
assembler [52]. Tools for polishing the resulting long read-only
assembly include Racon [53], Medaka [54], and Nanopolish [55];
Medaka is the latest entry into this rapidly growing and changing
field of algorithms and claims substantial accuracy and run time
improvements over Nanopolish.
3.6.3 Hybrid De Novo
Assembly
The primary disadvantage of using a singular sequencing technology is the necessary choice between either a highly accurate but
fragmented short read-only assembly or a highly complete but
relatively inaccurate long read-only assembly. When these technologies are used in conjunction with each other, the advantages of
each can be leveraged and their disadvantages mitigated, resulting
in a highly accurate and complete genome assembly. Such assemblies are called hybrid assemblies, since they combine more than
one type of sequencing technology. To date, Unicycler is generally
regarded as the best hybrid assembler for prokaryotic organisms
due to its optimizations at both the short-read assembly and longread assembly levels. Briefly, Unicycler begins with an optimized
SPAdes assembly and then proceeds to build “bridges” between
disconnected contigs using long-read data, both from a long-read
assembly and by mapping long reads directly onto the short-read
assembly. In a head-to-head benchmark comparison of Unicycler
versus other hybrid assemblers and assemblies using a variety of
sequencing technology, Unicycler was the clear top
performer [56].
416
Marco A. Riojas et al.