Assembly QC is broken down into three segments based on the 3 C's of assembly quality, assessing the Completeness, Contiguity, and Correctness of the assembly. Each script is run identically using QC_script.sh [Sample] [Assembler] [Mode] where the sample name is the same used for assembly and mode is either hic or kmer. QC outputs are present in the assembly directory ([Working_Directory]/[Sample]/Assemblies/[Assemblier]-[Mode]) under the corresponding 3 C's directory.
Most QC steps require the haplotypes outputed from the assembly step to be in .fa format. Use Utility-gfa2fa.sh [gfa] to convert them if you haven't already done so.
asmgene calculates the completeness of a genome by assesing the presence of coding sequences from a reference in a given assembly. asmgene is run using the default settings as specified in the documentation
QC_asmgene.sh [Sample] [Assembler] [Mode]
MashMap is used to assess assembly contiguity by comparing the assembly to a reference. Dotplots are used to visualise how the assembly compares to the reference, with the ideal situation being a single assemblied contig per reference chromosome. Multiple contigs per reference chromosome suggest gaps in the assembly that could not be scaffolded across.
QC_MashMap.sh [Sample] [Assembler] [Mode]
gfastats can be used to generate basic statistics about an assembly, including contiguity statistics N50, L50, etc.
QC_gfastats.sh [Sample] [Assembler] [Mode]
Contiguity can also be visualised using N50 plots, and can be used to compare contiguity across multiple haplotypes, assembliers, and individuals (note that the cross-individual comparisons are not currently support using the wrapper script but can be produced using the underlying python script). Note that this script uses .fai files produced from each of the assembly haplotypes to calculate N50, and can be produced using Utility-faidx.sh [fa]. The underlying python script for the N50 plots was written by Julian Lucas.
QC_N50Plot.sh [Sample.hap1.fai] [Sample.hap2.fai]
Assembly correctness can be calculated using the same k-mer tools as used for parental k-mer database creation, Yak and Meryl/Merqury. Both tools output a QV value, a log-scaled probability of error for the consensus base calls. QV above 50 (99.999% accuracy) is a 'good' quality to aim for.
Merqury correctness uses Meryl to produce a uncompressed 30 k-mer database for QV calculation.
QC_Merqury.sh [Sample] [Assembler] [Mode]
Yak correctness uses the same k-mer database creation as for assembly, so the can be created prior using the assembly script (Assembly_Yak.sh [Sample] [Sample]). If not created previous, the k-mer database will be created along with QV calculation using the below.
QC_Yak.sh [Sample] [Assembler] [Mode]