NanoBAC R package

The NanoBAC R package provides functions for the assembly of sequences from BAC or other large plasmids obtained using long read sequencing technologies such as Oxford Nanopore or Pacific Biosciences.

Installation

You can install the development version from GitHub with:

# install.packages("devtools")
  devtools::install_github("pgpmartin/NanoBAC")

A Singularity container with the NanoBAC package and its dependencies is also available (see definition on github)
First build the container nanobac.sifusing: singularity build nanobac.sif shub://pgpmartin/SingIMG:r36_nanobac

To open a shell inside the container, use: singularity shell nanobac.sif
To start the R version from the container, use: singularity exec nanobac.sif R
To run a script that uses the NanoBAC package, use: singularity exec nanobac.sif Rscript myRscript.R

Assembling repetitive regions of the genomes

Sequencing large plasmids using a long read technology is especially useful to obtain high quality sequences of repetitive regions, a prerequisite to identify small variations in these regions and assemble them in order to close the gaps that still remain in sequenced genomes. In particular, I developed the NanoBAC package while sequencing BACs containing repeats of the 45S ribosomal DNA (rDNA) gene of the plant model Arabidopsis thaliana. In the A. thaliana genome, 45S genes are organized as tandem repeats of ~10kb each and are located in 2 regions called Nucleolar Organizer Regions (NORs) located at the end of chromosome 2 and chromosome 4.

While short read sequencing platforms can identify some variations (such as SNPs or small indels << read length), they can’t reveal how these variations combine in individual rDNA gene copies or how these copies are organized along the NORs. Conversely, long reads from genomic DNA are still too noisy to reliably identify small variations.

Sequencing BACs, each containing several copies of rDNA genes or other types of repeats, using long read technologies allows to get the best of both worlds:

• obtaining multiple independent reads for each BAC and aligning them allows to correct the random errors inherent to these long reads and to gain high confidence in the identification of all variations
• long reads spanning several genes/repeats provide information on the combination of small variations present in each copy and on the organization of these copies relative to one another

By acquiring sequences for a number of long DNA fragments, it may be possible to assemble the whole NORs and thus close the biggest gap still present in the A. thaliana genome, and many other genomes as well. This is also the necessary step to gain a full understanding of how these rDNA copies are organized at the chromatin and 3D level and how their expression is regulated.

Long reads generated during sequencing

A plasmid or BAC is composed of:

• a vector (V) of known sequence
• a DNA insert (D) of unknown or partially known sequence

The DNA insert is cloned into the vector using a unique restriction/cloning site in the plasmid.

The different approaches to plasmid/BAC sequencing on long read platforms can generate different types of reads that are useful to assemble a high quality sequence:

• VDV reads start by a piece of the Vector sequence, followed by the full DNA insert sequence and end with a fragment of the Vector sequence
• DVD reads start by a fragment of the DNA insert sequence, followed by the full Vector sequence and end with a piece of the DNA insert sequence
• VD reads (or DV reads) contain a fragment (or the full) Vector sequence and a fragment (or the full) DNA insert sequence

We consider that the DNA insert sequence is unknown and repetitive. Thus, reads only containing the DNA insert sequence but no vector sequence (“D reads”) are generally not useful here because we don’t where to locate them with confidence along the repeats. In other words, the known and unique vector sequence plays the important role of anchoring the alignment between the reads so that we can correct the random sequencing errors and obtain a high quality assembly of the DNA insert.

Functions in the NanoBAC package

The functions in the NanoBAC package are used for the following tasks:

1. Create an interface between the alignment programs Blast/minimap2 and R:
   readBlast, blaST2GR, SelectSingularBlastALN, read_paf, paf2gr

2. Annotate the reads (especially in terms of read type: VDV, DVD, etc.):
   AnnotateBACreads, getVDVnames, getDVDnames, getVDnames

3. Select specific categories or reads / filtering the reads:
   FilterBACreads, estimateBACsizeFromVDV, selectVDVreads, splitDVDreads

4. identify the junctions between the vector sequence and the DNA insert sequence in VDV reads:
   findVDVjunctions

5. Obtain consensus sequences from the files generated by multiple sequence alignment softwares:
   consensusFromMSF, consmat2seq

In addition, the makeVarseq function allows to introduce random variations in a DNA sequence in order to create simulated reads.

These function are at the core of the NanoBAC pipeline. Below are examples of how to use the functions of the NanoBAC R package.
Start by loading the package:

library(NanoBAC)

Importing alignment results

In order to characterize the reads obtained during a sequencing run, it is necessary to align the vector sequence and potential known sequences that are expected in the DNA insert (e.g. the 18S rDNA sequence). This can be done efficiently using Blast which, in command line, has an option to output its result in a convenient tabular form: --outfmt=6.

We can import such data using: