Lineage-associated small inversions disrupt dosT, dnaE2, and a promoter-adjacent region in Mycobacterium tuberculosis

doi:10.21203/rs.3.rs-7319782/v1

Background: Large molecular inversions in the genome of Mycobacterium tuberculosis (Mtb) due to factors like the presence of insertion sequences and transposases are widely known. However, smaller inversions within CDS and noncoding control elements are rarely reported. The present study aims to identify inversions and their potential impact on Mtb biology in a lineage-specific manner.

Methodology: Structural variations (SVs) could only be detected by long reads. For this, we simulated long reads by de novo assembling the short read sequencing datasets and subsequently aligned representative strains from each lineage to the Progressive Mauve algorithm. Independently, long-read sequencing from the PacBio platform was acquired and analysed using SVIM. Variants were merged, and Fisher’s exact test was carried out to identify the inversion association with lineages. To visualize DNA features, the DNA-features-viewer tool was used.

Results:Simulated reads from short read sequencing gave indications of lineage-specific inversions. Long read sequencing approach led to identification of 7 unique inversions (2 associated with L1; 1 associated with L3, 2 associated with L4, and 2 associated with both L3 and L4, P<0.05). The inversions encompassed primarily non-essential genes like sdaA, dosT, Rv2026c, dnaE2, Rv1341, Rv1342, and lprD. An interesting inversion was observed in the upstream control element of purB and Rv0776c.

Conclusion: The study sheds light on small inversions that may be causing alterations in expression, formation of fusion genes, and nonsense mutations that may have a role in lineage-specific phenotypic changes.

Infectious Diseases

Bioinformatics

Epigenetics & Genomics

Inversions

Structural variations

TB

Fusion proteins

domain switching

Genetic inversions, a type of structural polymorphism, occur when a part of the genome is flipped in reverse orientation. In other words, when a region of the genome becomes reverse complementary relative to the rest of the genome, it is what is called a true inversion (1–3). Inversion-type structural variations are widely studied in organisms higher in the hierarchy, such as animals, plants, insects, and unicellular eukaryotes (4–6). In the case of bacteria, isolated literature suggests the presence of inversion events, such as rrnD-rrnE, which was observed in a laboratory strain of E. coli K12, and then the same inversion was reinverted years later. These events can potentially have a drastic impact on the phenotype of the bacteria, such as a growth disadvantage (7). Similarly, such events are reported in Clostridium and are mediated by recombinases encoded by RecV and are of larger sizes: 5-50 kbp (8). The inversions are also reported in promoter regions, altering expression levels, resistance development, and aiding adaptation in gut microbiota (9). Inversions in bacteria can present themselves within the coding region, leading to “domain switching” and alterations in the binding target proteins. Such events are reported in Bacteroides fragilis and Streptococcus pneumoniae (10–12). In Mycobacterium, owing to the presence of recombinases, large-scale intragenomic recombination events are inevitable, resulting in duplication, deletions, and inversions. In the event of duplication, the transposon can fit in the genome in either direction, potentially causing inversion (13,14). However, in the majority of Mtb-specific studies, large inversions caused by mobile genetic elements are reported (13,15,16).

The present study shows signatures of inversions through short-read sequencing data and actual seven small 197-681 bp inversions through the analysis of long-read sequencing datasets. Interestingly, despite being small, some of the inversions seem to disrupt more than one gene at a time. These inversions may have a role in lineage-specific prognosis differences.

Data acquisition: WGS data sets were downloaded from the European nucleotide archive in the form of fastq format. Some WGS data sets from the PACBIO ONT platform were not available for download in the FASTQ format from ENA. Such datasets were downloaded from the Sequence Read Archive (SRA) in SRA format. The files in SRA format were converted to fastq using the fastq-dump command from the SRA toolkit. The Bio Projects and lineage composition for short-read and long-read sequencing datasets are shown in Table S1.

Detecting Inversions using assemblies made using short-read sequencing: The short reads were first adapter-trimmed using bbduk.sh. Adapter-trimmed reads were subjected to Shovill (v1.1.0) (assembler: megahit) for de novo assembly and polishing (17,18). The assemblies were corrected using ragtag and subjected to WGS-based drug resistance and lineage detection using tb-profiler (v6.6.3) (19). The assemblies were subjected to ParsNP (v1.2) for the determination of distances (20). The tree from parsnp, the itol files from the tb-profiler output, were used to generate an annotated dendrogram. A phylogenetic free distance matrix was created, and 10 replicates per lineage were selected for analysis with ProgressiveMauve (v20150226) with and without the assumption of collinearity (21). Apart from 10 replicates per lineage, a representative complete Mycobacterium tuberculosis H37Rv genome (RefSeq: NC_000962.3) was also included in the Progressive Mauve alignment. The alignment output backbone format was parsed for the preparation of inversion and translocation matrices. The matrices were subjected to categorical PCA using the FactoMineR R package (22).

Detecting Inversions using long read WGS Mtb sequencing datasets: The single-ended long reads were filtered using seqkit (v2.3.0), and only reads with length > 500 bp were retained (23). The reads were subjected to de novo assembly using RAVEN (v1.6.1), and these were used for WGS-based drug resistance and lineage detection purposes with tb-profiler (19,24). Long sequencing reads were subjected to mapping with minimap2 (v2.26-r1175) using the reference genome (RefSeq: NC_000962.3) (25). The alignment files were sorted and subsequently sorted using samtools (26). The BAM files were used as input for SVIM (v2.0.0) for the detection of inversions (27). The sample ID and variants in VCF format were named and sorted using bcftools, sort and reheader options (v1.19) before collating into a multi-sample VCF. Survivor (v1.0.7) was used for merging single-sample VCF files into a single file (28). For merging, the maximum distance between breakpoints to be considered as a structural variant was set to 1000; the minimum presence of the inversion should be in at least 10 samples, distance-based SV size estimation and merging were enabled, and the minimum SV = 50 bp. The VCF file was converted to a table and subsequently parsed in MS-Excel and R for data analysis and visualizations.

Computational resources: All analyses were performed on a homegrown bioinformatic data analysis setup running WSL2 on AMD Ryzen 5600G with 64 GB RAM and another running on Intel i5-8265U with 24 GB RAM. The jobs were distributed and parallelized between the two computers using GNU parallel (v20160622). The tools used in the study are open-source and can be downloaded and installed from GitHub and Anaconda.

Statistical analysis: The FactomineR package was used for determining cPC components for each sample. For the small-scale association analysis, because of the small sample size, Fisher’s exact test was deemed appropriate and carried out in R. Benjamin Hochberg's correction of P values was also carried out. Inversions showing P<0.05 were considered significant and are reported in the present study.

Short read polished assemblies indicated the presence of inversions: From Illumina-based short sequencing reads deposited in SRA with BioProjects PRJEB41116 and PRJNA879962, a subset of 500 assemblies was selected. Randomization did not yield unique numbers; as a result, there were some duplicate assemblies that had to be removed. Finally, 467 WGS datasets were acquired from the European Nucleotide Archive (Table S1). Based on N50 (2817734±1888117), Genome fraction (98.9±11.5%), and total length (4403718±513257) from de novo assembly QC analysis, poor quality samples were identified and eliminated. WGS-based lineage detection was also correlated with samples and is included in the de novo assembly statistics table of short-read data sets, shown in Table S2. The subsequent analysis was done with a set of 467 pure lineage and high-quality samples. A schematic representation of procedure followed for analysis is shown in Figure 1A. Phylogenetic clustering of 467 samples showed lineage-specific clustering, and higher prevalence of drug resistance in lineage 2 isolates was observed (Figure 1B). A subset of 40 isolates (10 samples per lineage) was subsequently processed for alignment using the ProgressiveMauve algorithm. The corrected de novo assemblies that were aligned using ProgressiveMauve showed distorted patterns. Alignment-free clustering based on the mash distance matrix also clustered samples according to lineages. Again, lineage 2 samples from this subset showed resistance to a greater number of drugs, and only 2 of 10 lineage 2 isolates were sensitive (Figure 1C). Even though ProgressiveMauve results were not conclusive, the LCB matrices built using the approaches mentioned in the methodology section showed clustering on categorical PCA according to lineages as far as inversions were concerned (Figure 1D). Similar approach for translocations clustered samples into 4 subclusters. However, in the case of translocations, each cluster contained isolates from varying lineages (Figure 1E). This indicated that translocations are probably associated with an unknown factor, unlike in the case of lineages.

SV determination from long reads: Long-read WGS datasets of 237 PacBio datasets originating from cultures were identified using the metadata and downloaded. A subset (n=182) of datasets could be assembled successfully into fewer contigs. The others that could not be assembled were either due to poor quality (< 500 bp average read length) or limited computational resources. Lineage determination using tb-profiler revealed a major fraction of lineage 4 (n=113), followed by lineage 2 (n=26), lineage 1 (n=9), lineage 3 (n=6), and mixed lineages (n=28). The datasets representing mixed lineages were eliminated from subsequent analysis, except lineage2/4 samples. Inversion and BND were filtered for PCA analysis, which showed subtle clustering on the graph, indicating the presence of fewer INVs and BNDs between lineages (Figure 2A). Inversions with breakpoints on both strands were considered as true inversions, and those that had breakpoints only on one of the strands were, despite being significant, not considered true inversions. In Lineage 1, a set of two inversions, 652 bp and 310 bp, were observed with coordinates 76531-77183 (Raw P < 0.0001; BH P < 0.016) and 1482828-1483138 (Raw P < 0.01; BH P = 0.08), respectively (Figure 2B). In lineage 3, a set of three inversions of 639 bp, 681 bp and 539 bp were observed with coordinates 1508080-1508719 (Raw P = 0.01; BH P = 0.1); 2272338-2273019 (Raw P = 0.04; BH P = 0.1) and 3781352-3781891 (Raw P = 0.005; BH P = 0.09) (Figure 2C). Lineage 4 showed four inversions of 197 bp, 639 bp, 245 bp and 539 bp with coordinates 869783-869980 (Raw P = 0.02; BH; P = 0.2), 1508080-1508719 (Raw P = 0.008; BH P = 0.1), 2844929-2845174 (Raw P = 0.001; BH P = 0.02), and 3781352-3781891 (Raw P = 0.02; BH P = 0.2), respectively (Figure 2D). The observed lineage inversions in Lineage 1 of 652 bp and 310 bp were within the genes sdaA and Rv1320c (Figure 2E). Another inversion of 681 bp in lineage 3 was within but at the latter end of the coding region of dosT and within Rv2026c. Both these genes are reverse in direction. Probably this, disorientation, is causing a gene fusion or, more likely, disruption of both (Figure 2F). A set of two unique mutations that were not present in other lineages but found in lineage 4 included a 197 bp inversion in the upstream control region of the two genes oriented away from each other (Rv0776c and purB), probably having an impact on their expression. The other 245 bp inversion is within the centroid of the fas gene (Figure 2G). A set of two inversions was found, both in lineage 3 and lineage 4, that included a 539 bp inversion encompassing the latter end of Rv3369 and the latter end of dnaE2, and a 639 bp inversion disrupting three relatively 3 small genes oriented in opposite directions in lineages 3 and 4 was observed. The three genes included Rv1341 (forward), Rv1342 (reverse), and lprD (reverse) (Figure 2H). Some of the lineage 4 inversions were also found in samples of mixed lineages (lineage 2 and lineage 4), and these included 869783-869980, 1508080-1508719, and 2844929-2845174 (Figure S1). Due to low sample size, true inversions were not observed in lineage 2 isolates (Figure S1). A summary of inversions is included in Table 1.

Structural variations in coding and non-coding regions can have far-reaching impacts on bacterial phenotype. It is common to disrupt the genes of interest in laboratory conditions, and it is routinely done. However, in natural conditions, pathogenic bacteria may also be doing the same on their own through mechanisms already known or through yet-to-be-deciphered mechanisms. By disrupting some genes and reorienting promoter regions to express genes that favor Mtb’s survival, inversions can cause changes in host-pathogen interactions, growth defects, and virulence. To identify inversions in the Mtb genome, short-read WGS data were assembled into longer contigs to simulate long reads; however, this approach did not yield conclusive results matching earlier recommendations and literature that for the identification of structural variations, long-read sequencing is necessary. In lineage 1, we found two inversions, both in CDS-sdaA and Rv1320c. SdaA is a non-essential Iron-sulfur containing L-serine dehydratase and catalyzes serine deamination. Because of Inversion in sdaA, SdA levels might be low or have alternate functions due to potential domain switching. Lineage 1 is considered an ancient lineage and is considered to be less virulent compared to modern Mtb lineages (29). Rv1320c encodes adenylyl cyclase, which was found to have an inversion in its CDS, and is also known to be present in more virulent strains of Mtb (30). dosT encodes a two-component sensor histidine kinase DosT. It was also found to have inversion in its CDS. DosT has a role in sensing hypoxia and dormancy induction (31,32). So, an inversion in dosT CDS in lineage 3, particularly, might be responsible for dormancy differences that are well-known in Mtb lineages. The same inversion was also found to encompass its neighboring gene Rv2026c, which encodes for the Universal stress protein family protein and has a role in virulence, detoxification, and adaptation. Rv2026c is also known to get upregulated in stress conditions like oxidative stress. Rv2026c knockout strains are highly induced in hypoxic conditions, just like dosT (33). Probably, a cross-gene inversion between dosT and Rv3026c is leading to the generation of a fusion gene product with similar function. The role of Rv0776 is unknown, however it is under the control of a non-coding region that probably controls its expression, which was found to be inverted in lineage 3, and another gene that flanks this potential promoter region is purB, which encodes an RNAse-transformylase T, which has a role in the de novo purine salvage pathway (34). This reorientation of the promoter region flanked by two genes is interesting and requires more attention. In lineage 3, a centroid-region inversion was observed in fas gene that encodes fatty acid synthase Fas. The gene is relatively long, so a small inversion in its center may not be impacting its active site, as it is an essential gene. Its expression is known to be higher in strains other than lineages 2 and 4 (KZN605). The inversion might not have a role in expression level changes, but it might lead to the production of short-length proteins, which could not be detected, leading to the assumption that it has higher expression in Lineage 2 and KZN605 (35). dnaE2 encodes DNA polymerase III (alpha chain) DnaE2 (DNA nucleotidyltransferase), so an inversion in the latter end might not significantly impact its DNA polymerase activity. It is important to note that DnaE2 is an error-prone DNA polymerase and is upregulated in response to drugs, leading to mutations in drug target genes, causing drug resistance (36). An inversion in the latter end of dnaE2 of lineage 3 could be of interest. Another inversion found in both lineage 3 and 4 encompassed 3 genes oriented in different directions: Rv1341 (encodes a conserved protein), Rv1342c (membrane protein), and lprD (encodes lipoprotein). The inversion might be causing an alteration in the expression of Rv1342c and leading to the generation of fusion genes of lprD and Rv1341.

The study sheds light on short 197-681 bp long inversions in CDS and non-coding regions that could have an impact on phenotypes, lead to the formation of fusion gene products, and alter expression in a lineage-specific manner. Further experimental investigation is necessary to validate these findings.

Data availability statement: All data generated are included in the manuscript. The study used publicly available WGS datasets in NCBI-SRA. The WGS datasets used in the study are provided in the supplementary material with BioProjects accession IDs in Table S1.

Ethics and Consent: Publicly available M. tuberculosis WGS datasets are used in the study. There was no involvement of aspects that required ethical clearance or consent.

Contributions: NB contributed to: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Visualization, Writing – original draft, Writing – review & editing.

Competing interests: There are no conflicting interests (Single author-driven study).

Grant information: The work presented was carried out of curiosity, and no formal funding was received.

Acknowledgement: NB thanks the authors who developed data analysis tools and made them open source, and the authors of the WGS datasets used in this study.

Chaisson MJP, Sanders AD, Zhao X, Malhotra A, Porubsky D, Rausch T, et al. Multi-platform discovery of haplotype-resolved structural variation in human genomes. Nat Commun. 2019 Apr 16;10:1784.
Zhang W, Tan C, Hu H, Pan R, Xiao Y, Ouyang K, et al. Genome architecture and diverged selection shaping pattern of genomic differentiation in wild barley. Plant Biotechnol J. 2022 Nov 16;21(1):46–62.
West PT, Chanin RB, Bhatt AS. From genome structure to function: insights into structural variation in microbiology. Curr Opin Microbiol. 2022 Oct;69:102192.
Ayala D, Ullastres A, González J. Adaptation through chromosomal inversions in Anopheles. Front Genet. 2014 May 21;5:129.
Coghlan A, Eichler EE, Oliver SG, Paterson AH, Stein L. Chromosome evolution in eukaryotes: a multi-kingdom perspective. Trends Genet. 2005 Dec;21(12):673–82.
Hewitt SK, Donaldson I, Lovell SC, Delneri D. Sequencing and characterisation of rearrangements in three S. pastorianus strains reveals the presence of chimeric genes and gives evidence of breakpoint reuse. PLoS ONE. 2014 Mar 18;9(3):e92203.
Hill CW, Gray JA. Effects of Chromosomal Inversion on Cell Fitness in Escherichia Coli K-12. Genetics. 1988 Aug;119(4):771–8.
Sekulovic O, Mathias Garrett E, Bourgeois J, Tamayo R, Shen A, Camilli A. Genome-wide detection of conservative site-specific recombination in bacteria. PLoS Genet. 2018 Apr 5;14(4):e1007332.
Jiang X, Hall AB, Arthur TD, Plichta DR, Covington CT, Poyet M, et al. Invertible promoters mediate bacterial phase variation, antibiotic resistance, and host adaptation in the gut. Science. 2019 Jan 11;363(6423):181–7.
Nakayama-Imaohji H, Hirakawa H, Ichimura M, Wakimoto S, Kuhara S, Hayashi T, et al. Identification of the Site-Specific DNA Invertase Responsible for the Phase Variation of SusC/SusD Family Outer Membrane Proteins in Bacteroides fragilis. J Bacteriol. 2009 Oct;191(19):6003–11.
Kuwahara T, Yamashita A, Hirakawa H, Nakayama H, Toh H, Okada N, et al. Genomic analysis of Bacteroides fragilis reveals extensive DNA inversions regulating cell surface adaptation. Proc Natl Acad Sci U S A. 2004 Oct 12;101(41):14919–24.
Li J, Li JW, Feng Z, Wang J, An H, Liu Y, et al. Epigenetic Switch Driven by DNA Inversions Dictates Phase Variation in Streptococcus pneumoniae. PLoS Pathog. 2016 July 18;12(7):e1005762.
Ruesen C, Chaidir L, van Laarhoven A, Dian S, Ganiem AR, Nebenzahl-Guimaraes H, et al. Large-scale genomic analysis shows association between homoplastic genetic variation in Mycobacterium tuberculosis genes and meningeal or pulmonary tuberculosis. BMC Genomics. 2018 Feb 5;19:122.
Bhalla N, Nanda RK. Pangenome-wide association study reveals the selective absence of CRISPR genes (Rv2816c-19c) in drug-resistant Mycobacterium tuberculosis. Microbiol Spectr. 12(8):e00527-24.
Morgado SM, Paulo Vicente AC. Genomics of Atlantic Forest Mycobacteriaceae strains unravels a mobilome diversity with a novel integrative conjugative element and plasmids harbouring T7SS. Microb Genomics. 2020;6(7):e000382.
Shitikov EA, Bespyatykh JA, Ischenko DS, Alexeev DG, Karpova IY, Kostryukova ES, et al. Unusual Large-Scale Chromosomal Rearrangements in Mycobacterium tuberculosis Beijing B0/W148 Cluster Isolates. PLOS ONE. 2014 Jan 8;9(1):e84971.
Li D, Liu CM, Luo R, Sadakane K, Lam TW. MEGAHIT: an ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinforma Oxf Engl. 2015 May 15;31(10):1674–6.
Seemann T. tseemann/shovill [Internet]. 2025 [cited 2025 July 21]. Available from: https://github.com/tseemann/shovill
Phelan JE, O’Sullivan DM, Machado D, Ramos J, Oppong YEA, Campino S, et al. Integrating informatics tools and portable sequencing technology for rapid detection of resistance to anti-tuberculous drugs. Genome Med. 2019 June 24;11(1):41.
Kille B, Nute MG, Huang V, Kim E, Phillippy AM, Treangen TJ. Parsnp 2.0: scalable core-genome alignment for massive microbial datasets. Bioinforma Oxf Engl. 2024 May 2;40(5):btae311.
Darling AE, Mau B, Perna NT. progressiveMauve: multiple genome alignment with gene gain, loss and rearrangement. PloS One. 2010 June 25;5(6):e11147.
Lê S, Josse J, Husson F. FactoMineR: An R Package for Multivariate Analysis. J Stat Softw. 2008 Mar 18;25:1–18.
Shen W, Le S, Li Y, Hu F. SeqKit: A Cross-Platform and Ultrafast Toolkit for FASTA/Q File Manipulation. PloS One. 2016;11(10):e0163962.
Vaser R, Šikić M. Time- and memory-efficient genome assembly with Raven. Nat Comput Sci. 2021 May;1(5):332–6.
Li H. Minimap2: pairwise alignment for nucleotide sequences. Bioinforma Oxf Engl. 2018 Sept 15;34(18):3094–100.
Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinforma Oxf Engl. 2009 Aug 15;25(16):2078–9.
Heller D, Vingron M. SVIM: structural variant identification using mapped long reads. Bioinforma Oxf Engl. 2019 Sept 1;35(17):2907–15.
Sedlazeck F. fritzsedlazeck/SURVIVOR [Internet]. 2025 [cited 2025 July 21]. Available from: https://github.com/fritzsedlazeck/SURVIVOR
Romagnoli A, Petruccioli E, Palucci I, Camassa S, Carata E, Petrone L, et al. Clinical isolates of the modern Mycobacterium tuberculosis lineage 4 evade host defense in human macrophages through eluding IL-1β-induced autophagy. Cell Death Dis. 2018 May 24;9(6):624.
Rindi L, Lari N, Garzelli C. Genes of Mycobacterium tuberculosis H37Rv downregulated in the attenuated strain H37Ra are restricted to M. tuberculosis complex species. New Microbiol. 2001 July;24(3):289–94.
Honaker RW, Leistikow RL, Bartek IL, Voskuil MI. Unique roles of DosT and DosS in DosR regulon induction and Mycobacterium tuberculosis dormancy. Infect Immun. 2009 Aug;77(8):3258–63.
Kumar A, Toledo JC, Patel RP, Lancaster JR, Steyn AJC. Mycobacterium tuberculosis DosS is a redox sensor and DosT is a hypoxia sensor. Proc Natl Acad Sci U S A. 2007 July 10;104(28):11568–73.
Hingley-Wilson SM, Lougheed KEA, Ferguson K, Leiva S, Williams HD. Individual Mycobacterium tuberculosis universal stress protein homologues are dispensable in vitro. Tuberc Edinb Scotl. 2010 July;90(4):236–44.
Chen C, Liu Z, Liu L, Wang J, Jin Q. Structural characterization of glycinamide-RNase-transformylase T from Mycobacterium tuberculosis. Emerg Microbes Infect. 2020;9(1):58–66.
Moopanar K, Nyide ANG, Senzani S, Mvubu NE. Clinical strains of Mycobacterium tuberculosis exhibit differential lipid metabolism-associated transcriptome changes in in vitro cholesterol and infection models. Pathog Dis. 2022 Dec 12;81:ftac046.
Salini S, Bhat SG, Naz S, Natesh R, Kumar RA, Nandicoori VK, et al. The Error-Prone Polymerase DnaE2 Mediates the Evolution of Antibiotic Resistance in Persister Mycobacterial Cells. Antimicrob Agents Chemother. 2022 Mar 15;66(3):e01773-21.

Table 1: Summary of statistically significant inversions. Inversions encompassed CDS and control elements disrupting or potentially altering their expression, respectively. P values were determined using Fisher's exact test to analyze their lineage associations. Abbreviations: BH: Benjamin Hochberg.

S. No.	Lineage	Start	End	Raw P	BH P-value	Genomic region
1	Lineage 1	76531	77183	1.52E-04	1.69E-02	sdaA (Rv0069c)
2	Lineage 1	1482828	1483138	6.68E-03	8.03E-02	Rv1320c
3	Lineage 3	1508080	1508719	1.54E-02	1.40E-01	Rv1341, Rv1342c, IprD (Rv1343c)
4	Lineage 3	2272338	2273019	4.09E-02	1.87E-01	Rv2026c, dosT (Rv2027c)
5	Lineage 3	3781352	3781891	5.38E-03	9.92E-02	Rv3369, dnaE2 (Rv3370c)
6	Lineage 4	869783	869980	2.79E-02	2.29E-01	CDS flanked by Rv0776c (upstream) and purB (Rv0777) (Downstream)
7	Lineage 4	1508080	1508719	8.46E-03	1.19E-01	Rv1341, Rv1342c, IprD (Rv1343c)
8	Lineage 4	2844929	2845174	1.03E-03	2.91E-02	fas (Rv2524c)
9	Lineage 4	3781352	3781891	2.71E-02	2.26E-01	Rv3369, dnaE2 (Rv3370c)

The authors declare no competing interests.

Extendeddata.pdf
Supplementary

Lineage-associated small inversions disrupt dosT, dnaE2, and a promoter-adjacent region in Mycobacterium tuberculosis

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Discussion

Declarations

References

Table

Additional Declarations

Supplementary Files

Status:

Version 1