Pan-lineage Mycobacterium tuberculosis reference genome for enhanced molecular diagnosis.

In Mycobacterium tuberculosis (MTB) control, whole genome sequencing-based molecular drug susceptibility testing (molDST-WGS) has emerged as a pivotal tool. However, the current reliance on a single-strain reference limits molDST-WGS's true potential. To address this, we introduce a new pan-lineage reference genome, 'MtbRf'. We assembled 'unmapped' reads from 3,614 MTB genomes (751 L1; 881 L2; 1,700 L3; and 282 L4) into 35 shared, annotated contigs (54 coding sequences [CDSs]). We constructed MtbRf through: (1) searching for contig homologues among genome database that precipitate results uniquely within Mycobacteria genus; (2) comparing genomes with H37Rv ('lift-over') to define 18 insertions; and (3) filling gaps in H37Rv with insertions. MtbRf adds 1.18% sequences to H37rv, salvaging >60% of previously unmapped reads. Transcriptomics confirmed gene expression of new CDSs. The new variants provided a moderate DST predictive value (AUROC 0.60-0.75). MtbRf thus unveils previously hidden genomic information and lays the foundation for lineage-specific molDST-WGS.

SigAlign: an alignment algorithm guided by explicit similarity criteria.

In biological sequence alignment, prevailing heuristic aligners achieve high-throughput by several approximation techniques, but at the cost of sacrificing the clarity of output criteria and creating complex parameter spaces. To surmount these challenges, we introduce 'SigAlign', a novel alignment algorithm that employs two explicit cutoffs for the results: minimum length and maximum penalty per length, alongside three affine gap penalties. Comparative analyses of SigAlign against leading database search tools (BLASTn, MMseqs2) and read mappers (BWA-MEM, bowtie2, HISAT2, minimap2) highlight its performance in read mapping and database searches. Our research demonstrates that SigAlign not only provides high sensitivity with a non-heuristic approach, but also surpasses the throughput of existing heuristic aligners, particularly for high-accuracy reads or genomes with few repetitive regions. As an open-source library, SigAlign is poised to become a foundational component to provide a transparent and customizable alignment process to new analytical algorithms, tools and pipelines in bioinformatics.

A High Quality Asian Genome Assembly Identifies Features of Common Missing Regions.

The current human reference genome (GRCh38), with its superior quality, has contributed significantly to genome analysis. However, GRCh38 may still underrepresent the ethnic genome, specifically for Asians, though exactly what we are missing is still elusive. Here, we juxtaposed GRCh38 with a high-contiguity genome assembly of one Korean (AK1) to show that a part of AK1 genome is missing in GRCh38 and that the missing regions harbored ~1390 putative coding elements. Furthermore, we found that multiple populations shared some certain parts in the missing genome when we analyzed the "unmapped" (to GRCh38) reads of fourteen individuals (five East-Asians, four Europeans, and five Africans), amounting to ~5.3 Mb (~0.2% of AK1) of the total genomic regions. The recovered AK1 regions from the "unmapped reads", which were the estimated missing regions that did not exist in GRCh38, harbored candidate coding elements. We verified that most of the common (shared by ≥7 individuals) missing regions exist in human and chimpanzee DNA. Moreover, we further identified the occurrence mechanism and ethnic heterogeneity as well as the presence of the common missing regions. This study illuminates a potential advantage of using a pangenome reference and brings up the need for further investigations on the various features of regions globally missed in GRCh38.