There are several programs that can operate with ONT reads. In this workshop, we will rely on canu, which is a program designed to perform ONT read correction, filtering (removal of bases or fragments from sequences with low quality) and scaffold assembly. The program requires a lot of resources for its execution (this depends on the size of the genomes, their complexity and the initial amount of data being used for its execution. The program recommends starting the run with an amount of data that represents between 20x-60x+ theoretical coverage (e.g., Sun et al., 2021). The program can operate with lower ammounts of input data but at the expense of higher error rates (higher number of miscalled bases on the read after correction analysis).
Excerpt from Sun et al.,2021
The following assemblers were used for the benchmarking, including the long-read only assemblers (Canu [12], Flye [13], Wtdbg2 [14], Miniasm [15], NextDenovo >>(https://github.com/Nextomics/NextDenovo), NECAT [6], Raven [16] and Shasta [7]) and hybrid assemblers (MaSuRCA [17] and QuickMerge [18]). Canu was not tested on the M. coruscus genome owing to the extremely intensive computing time required for this large genome. Previous analyses have suggested that using corrected ONT reads could improve the genome assembly [19]. To check the effect that this has on the assemblies, the ONT reads that were corrected and/or trimmed by Canu.
canu is a modular program, this means that each mode (correction, trim, assembly) can be run separately. This is recommended as in some instances the program may crash due to lack of resources, especially when assembling complex genomes and using lots of input data (e.g., nuclear genomes of +1 Gb). In this case, it is recommended to use canu for corrections and filtering, and other programs for assembling corrected and filtered data (e.g. SMARTdenovo).
In this section of the workshop, we will correct, trim and assemble a small subset of the Artocarpus altilis dataset sequenced from a sample collected in a dry environment (WGS17). This will enable is to use canu for correction, trimming and assembly. We will aim to assemble the plastid genome of A. altilis!
Important
The data required to run this section should be made available in the folder NGSdat/fastq_subset/ of your local directory. These are *.fastq files contain reads that match the plastid genome. These files are available here. Make sure to download it and store in the NGSdat/canu_fastq directory in your local machine (have a look at the README.md file in the folder canu_fastq folder of this repo).
To run the program in correction mode, use:
canu -correct genomeSize=160k -nanopore-raw *.fastq -p canu_corr -d ../canu_corr/Where:
genomeSize # indicates the size of the genome of the organism to be analysed (in Mb, Gb, or Kb)
-nanopore-raw # indicates what kind of inputs are being used
-p # prefix of output files
-d # output directoryWe will use here a genome size of 160,000 bp because is the standard size of a plastid genome (also, a plastid genome of reference of A. altilis, available in GenBank, indicates that it has 160,184 bp).
This command will output reads that have been corrected. Correction occurres by building overlap graphics (see Fig. 2 of this paper). This commmand will also issue a report that looks like this:
[CORRECTION/LAYOUT]
-- original original
-- raw reads raw reads
-- category w/overlaps w/o/overlaps
-- -------------------- ------------- -------------
-- Number of Reads 6030 544
-- Number of Bases 33156409 1831339
-- Coverage 207.228 11.446
-- Median 4152 2563
-- Mean 5498 3366
-- N50 6467 4381
-- Minimum 1001 0
-- Maximum 62276 41616
--
-- --------corrected--------- ----------rescued----------
-- evidence expected expected
-- category reads raw corrected raw corrected
-- -------------------- ------------- ------------- ------------- ------------- -------------
-- Number of Reads 6026 491 491 133 133
-- Number of Bases 31502614 6839053 6400196 1090108 291738
-- Coverage 196.891 42.744 40.001 6.813 1.823
-- Median 4036 11603 11237 5348 1914
-- Mean 5227 13928 13035 8196 2193
-- N50 6056 13582 12560 12933 2395
-- Minimum 1001 8831 8794 1523 1036
-- Maximum 56138 56138 38973 33858 6024
--
-- --------uncorrected--------
-- expected
-- category raw corrected
-- -------------------- ------------- -------------
-- Number of Reads 5950 5950
-- Number of Bases 27058587 14425376
-- Coverage 169.116 90.159
-- Median 3787 1928
-- Mean 4547 2424
-- N50 5166 4992
-- Minimum 0 0
-- Maximum 62276 48522
--
-- Maximum Memory 1083529106To run canu in filter mode (trim), use:
canu -trim genomeSize=160k -nanopore-corrected canu_corr.correctedReads.fasta.gz -p canu_trim -d ../canu_trim/Where:
-nanopore-corrected # indicates what type of inputs are being used
-p # prefix of the output files
-d # output directoryThis command will output reads that have been trimmed.
Finally, assembling genomes with corrected ONT data can be more or less staright forward depending on the complexity of the genome (ploidy levels, proportion of repetitive elements).
To assemble the plastid genome using canu and trimmed reads, try the following command:
canu -assemble -p A_altilis_CP genomeSize=160k correctedErrorRate=0.14 -nanopore-corrected canu_trim.trimmedReads.fasta.gz -d ../canu_ass/Where:
-nanopore-corrected # indicates what type of inputs are being used
correctedErrorRate=0.14 # indicates the overalp error rate. The value here is adequate for corrected nanopore reads.
-d # output directoryIn some instances, using canu for genome assembly might become prohibited when lots of input data are being used, or complex, large genomes are being assembled. In such case, a good alternative for conducting genome assembly is SMARTdenovo. This program runs considerably much faster than canu, with a comparable performance. It requires as input data raw or corrected ONT reads.
To run SMARTdenovo, use the following command.
smartdenovo.pl -c 1 -t 64 -p prefix_output *fastq.fastq > prefix_output.mak
make -f *.makWhere:
-p # output prefix
-t # number of threads (default is 8)
-k # k-mer length for overlapping (default is 16)
-J # min read length (default is 5000)
-c # generate consensus sequence- Attempt to assemble a draft genome from the trimmed
canureads, using SMARTdenovo. How different is the resulting genome assemnbly from that produced bycanu?
When assembling genomes with ONT data, it is extremely important to ensure that the scaffolds are as accurate as possible. Here, erroneously sequenced based from ONT data might have remained could prevent better contiguity in the assembly. One way to improve accuracy is by "polishing" the genome, using short read sequencing data produced from the exact same individual. This is done in two steps.
- Mapping the illumina sequencing data against the genome assembly, using read aligners, like
minimap2. This programs generates an index files with coordinates of the reads mapped against the genome assembly. Here, every read has a quality mapping score, and information on on when bases from the short reads disagree with the reference. THe ouput of this command is a SAM file (read here more about SAM files). In this workshop, We wont run these commands because we lack the illumina sequencing data. To runminimap2, execute:
minimap2 -ax sr -t 64 ../genome.assembly.fasta input_illumina_read_files_R1_001.fastq input_illumina_read_files_R2_002.fastq > output_minimap.samWhere:
-ax # map short reads.
-t # number of threads- Correct the scaffolds using the SAM file produced by
minimap2, and the ONT filtered and corrected reads (produced bycanu). We useraconto this end.
racon -t 64 corrected.trimmed.canu.reads.fasta output_minimap.sam genome.assembly.smartdenovo.fastaWhere:
-t # number of threads- Koren S, Walenz BP, Berlin K, Miller J, Bergman NH, Phillippy A. (2017) Canu: scalable and accurate long-read assembly via adaptive k-mer weighting and repeat separation. Genome Res. 27: 722-736
- Sun J, Li R, Chen C, Sigwart JD, Kocot KM. (2021) Benchmarking ONT read assemblers for high-quality molluscan genomes. Proc. B Phil Trans. https://doi.org/10.1098/rstb.2020.0160
- Liu H, Wu S, Li A, Ruan J. (2021). SMARTdenovo: a de-novo assembler using long noisy reads. Gigabyte. https://gigabytejournal.com/articles/15
- Vaser R, Sovic I, Nagarajan N, Sikic M. (2017). Fast and accurate de novo genome assembly from long uncorrected reads. Genome Res. https://genome.cshlp.org/content/early/2017/01/18/gr.214270.116
- Li H. (2018). Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics, 34:3094-3100. https://academic.oup.com/bioinformatics/article/34/18/3094/4994778?login=true