MAC: Merging Assemblies by Using Adjacency Algebraic Model and Classification

Abstract

With the generation of a large amount of sequencing data, different assemblers have emerged to perform de novo genome assembly. As a single strategy is hard to fit various biases of datasets, none of these tools outperforms the others on all species. The process of assembly reconciliation is to merge multiple assemblies and generate a high-quality consensus assembly. Several assembly reconciliation tools have been proposed. However, the existing reconciliation tools cannot produce a merged assembly which has better contiguity and contains less errors simultaneously, and the results of these tools usually depend on the ranking of input assemblies. In this study, we propose a novel assembly reconciliation tool MAC, which merges assemblies by using the adjacency algebraic model and classification. In order to solve the problem of uneven sequencing depth and sequencing errors, MAC identifies consensus blocks between contig sets to construct an adjacency graph. To solve the problem of repetitive region, MAC employs classification to optimize the adjacency algebraic model. What's more, MAC designs an overall scoring function to solve the problem of unknown ranking of input assembly sets. The experimental results from four species of GAGE-B demonstrate that MAC outperforms other assembly reconciliation tools.

High-resolution shotgun metagenomics: the more data, the better?

“…The execution of such a workflow would however require an accepted computational method to merge or combine multiple assemblies together, which to the best of our knowledge does not exist. Software reported to perform this type of merging task are usually not maintained anymore and not suitable for modern large metagenomic assemblies [14][15][16] and target single genome assemblies [17][18][19] . At first glance, long reads (PacBio, Oxford Nanopores) can also seem attractive to replace short-reads in the objective of obtaining more contiguous co-assemblies, but performing a multi-library co-assembly of long reads data type also requires enormous amounts of RAM and compute time (personal observations), especially if reads need to be corrected prior to be assembled, as it is often the case with error-prone long reads data types.…”

Structural insights into coordinating 5S RNP rotation with ITS2 pre‐RNA processing during ribosome formation

“…Using the wild‐type reads, we assembled a draft genome with SPAdes v3.12.0 (Bankevich et al , 2012) (295 contigs, N50 729,905 bp, sum 28.41 Mbp) and in a second step, we combined the draft genome with the current reference genome with a tool called MAC2.0 modifying the length option of show‐coords to “‐L 10000” (Tang et al , 2019). After gap closing with GapCloser v1.12 (Luo et al , 2012), only 2 gaps remained in the mitochondrial chromosome.…”

Rapid transcriptomic and physiological changes in the freshwater pennate diatomMayamaea pseudoterrestrisin response to copper exposure

“…Both assemblies were merged with Raven-based assembly (query) and NextDenovo-based assembly (reference) using MAC 2.0. 34 The merged assembly was polished using POLCA 35 in MaSuRCA version 4.0.4, 36 using trimmed Illumina reads. Subsequently, because diatoms are diploid, haplotigs were purged from the merged assembly using Purge Haplotigs version 1.1.1, 37 with the following options: -l 5, -m 25, and -h 80.…”