Genome assembly quality control.
Anthony Bretaudeau
Alexandre Cormier
Stéphanie Robin
Erwan Corre
Laura Leroi
Christophe Klopp
Requirements
Before diving into this slide deck, we recommend you to have a look at:
- Sequence analysis
- Quality Control: slides slides - tutorial hands-on
N50 & L50
N50: given a set of sequences of varying lengths, the N50 is defined as the length L of the shortest contig for which longer and equal length contigs cover at least 50% of the assembly.
L50: given a set of sequences of varying lengths, the L50 is defined as count of smallest number of sequences whose length sum makes up 50% of the assembly.
N50 describes a sequence length whereas L50 describes a number of sequences.
N50 & L50
Example:
- Genome size = 100
- Sequence sorted by size list L = (25, 10, 10, 8 , 7, 7 , 6 , 5, 5, 5, 5, 3, 2, 2 ) = 100
- 50% of the total length is contained within sequences of at least 8bp: 25 + 10 + 10 + 8 ≥ 50
N50 = 8 and L50 = 4
Alhakami, H., Mirebrahim, H., & Lonardi, S. (2017). A comparative evaluation of genome assembly reconciliation tools. Genome biology, 18(1), 1-14.
QUAST - A tool to evaluate genome assemblies
- QUAST: for genome assemblies.
- MetaQUAST: for metagenomic datasets.
- QUAST-LG: for large genomes (e.g., mammalians).
- rnaQUAST: for RNAseq.
- Icarus: an interactive visualizer for these tools.
It also includes:
- Reads mapping (mi-assemblies evaluation).
- Kmer representation (KMC)
- Structural prediction modules (GeneMark, GlimmerHMM, Barrnap and BUSCO).
- For metagenomics dataset: MetaGeneMark, Krona tools, BLAST, and SILVA 16S rRNA database.
Core genes evaluation software
BUSCO: Assessing genome assembly and annotation completeness with Benchmarking Universal Single-Copy Orthologs
Eukaryota: 255 single copy from 70 species; Arthropoda: 1013 single copy from 90 species; Fungi: 758 single copy from 549 species
Waterhouse, R. M., Zdobnov, E. M. & Kriventseva, E. V. Correlating Traits of Gene Retention, Sequence Divergence, Duplicability and Essentiality in Vertebrates, Arthropods, and Fungi. Genome Biol Evol 3, 75–86 (2011).
Assembly kmer content
The aim is to check assembly coherence against the content within reads that were used to produce the assembly. Basically, how many elements of each frequency on the read’s spectrum ended up being not included in the assembly, included once, included twice etc.
- Merqury or KAT
- Histogram is build with read kmer content.
Reads mapping and assembly coverage
- Proportion of mapped vs. unmapped reads i.e. proportion of missing parts in the assembly
- Coverage in terms of redundancy (A): number of reads that align to, or "cover," a known reference.
- Coverage in terms of the percentage coverage of a reference by reads (B): E.g. if 90% of a reference is covered by reads (and 10% not) it is a 90% coverage.
Switch and hamming errors (phased assemblies)
- Switch error: a change from one parental allele to another parental allele on a contig. This terminology has been used for measuring reference-based phasing accuracy for two decades. A haplotig is supposed to have no switch errors.
- Yak hamming error: an allele not on the most supported haplotype of a contig. Its main purpose is to test how close a contig is to a haplotig. The yak definition is not widely accepted. The hamming error rate is arguably less important in practice.
http://lh3.github.io/2021/04/17/concepts-in-phased-assemblies
Tips
The quality of an assembly is often validated by using other data from the same individual or from other individuals (RNA-Seq alignment, Hi-C alignment, DNA-Seq alignment,...).
The positions of the telomeric repeats in the chromosome assemblies are also of interesting to evaluate the correctness.
The identification of organelles (mitochondria, chloroplast,...) can also inform us about the quality of the assembly in terms of completness. However, the structure of the organelles may lead the assembler to think that they are repeats and he discards them.
In the case of diploid organisms, one of the classical problems of assemblies is the conservation of the two haplotypes. We obtains particular BUSCO / kmer / assembly size metrics that can be corrected by removing, "purging", the haplotigs.
Thank You!
This material is the result of a collaborative work. Thanks to the Galaxy Training Network and all the contributors!
This material is licensed under the Creative Commons Attribution 4.0 International License.