Supplemental Figures Supplemental Figure 1.

Size: px

Start display at page:

Download "Supplemental Figures Supplemental Figure 1."

Stanley Lamb
6 years ago
Views:

Supplemental Material: Annu. Rev. Genom. Hum. Genet. 2017. 18:321-356 https://doi.org/10.1146/annurev-genom-091416-035526 A Robust Framework for Microbial Archaeology Warinner et al.

1 Supplemental Material: Annu. Rev. Genom. Hum. Genet : A Robust Framework for Microbial Archaeology Warinner et al. Supplemental Figures Supplemental Figure 1. Mismapping of environmental (soil and ocean) and host-associated (saliva) metagenomic sequences to selected obligate pathogens. Contaminants can produce false positive signals of human pathogens even at relatively low sequencing depths; in addition to mapping, further validation is necessary to confirm the identification of a pathogen from metagenomic data. Soil, ocean, and saliva metagenomes were obtained from the EBI-ENA and NCBI-SRA databases and pre-processed with the EAGER pipeline (13) using the ClipAndMerge module. The merged reads were mapped to fourteen pathogen genomes using BWA (aln default settings) (9). Mapping statistics were obtained from the ReportTable module of EAGER and normalized to report the number of mapped hits per million reads. See Supplemental Appendix for further details.

The alignment was performed with MALT (4) using default parameters to a database consisting of complete bacterial genomes. Alignments to M.

2 Supplemental Figure 2. Ancient DNA data from published studies exhibiting the patterns described in Figure 5. (a) Genomic coverage shown for shotgun DNA sequencing data from a Hungarian mummy lung sample (2) aligned to the M. tuberculosis reference genome. The alignment was performed with MALT (4) using default parameters to a database consisting of complete bacterial genomes. Alignments to M. tuberculosis are visualized using the alignment viewer of MEGAN6 (5). Aligned reads are distributed evenly across the whole genome. (b) Genomic coverage shown for shotgun DNA sequencing data from a Spanish Chalcolithic dentine sample (19) aligned to the M. tuberculosis reference genome. The alignment was performed with MALT using default parameters to a database consisting of complete bacterial genomes. Alignments to M. tuberculosis are visualized using the alignment viewer of MEGAN6. Thin grey vertical lines indicate genomic regions with no coverage, which have been collapsed. Many reads are accumulating at distinct loci instead of being distributed randomly. (c) Histogram showing the distribution of percent identity values of shotgun DNA sequencing reads from a Hungarian mummy lung sample (2) aligned to the M. tuberculosis reference genome. The alignment was performed with MALT using default parameters to a database consisting of complete bacterial genomes. Most reads show very high percent identity values close to 100%. (d) Histogram showing the distribution of percent identity values of shotgun DNA sequencing reads from a

3 Spanish Chalcolithic dentine sample (19) aligned to the M. tuberculosis reference genome. The alignment was performed with MALT using default parameters to a database consisting of complete bacterial genomes. Most reads are rather dissimilar to the reference with percent identity values below 95%. (e) Histogram showing the distribution of SNP allele frequencies after SNP calling based on an alignment of tuberculosis-enriched DNA sequencing data from a tuberculosis positive sample from ancient South American human remains (54U) (1) to the M. tuberculosis reference genome. Reads were aligned with BWA (9) using strict mapping parameters (see (1) for details). Only multiallelic sites are plotted. Only sporadic multiallelic calls can be observed. (f) Histogram showing the distribution of SNP allele frequencies after SNP calling based on an alignment of shotgun DNA sequencing data from a Hungarian mummy lung sample (2) to the M. tuberculosis reference genome. Reads were aligned with BWA using strict mapping parameters (see (1) for details). Only multiallelic sites are plotted. A symmetric distribution of SNP allele frequencies around 50% indicates the presence of two different strains in equal abundance. (g) Histogram showing the distribution of SNP allele frequencies after SNP calling based on an alignment of shotgun DNA sequencing data from a modern tuberculosis sample (1) to the M. tuberculosis reference genome. Reads were aligned with BWA using strict mapping parameters (see (1) for details). Only multiallelic sites are plotted. An asymmetric distribution of SNP allele frequencies indicates the presence of two different strains in unequal abundance. (h) Histogram showing the distribution of SNP allele frequencies after SNP calling based on an alignment of tuberculosis-enriched DNA sequencing data from a tuberculosis positive sample from ancient South American human remains (54U) (1) to the M. tuberculosis reference genome. Reads were aligned with BWA using default mapping parameters (see (1) for details). Only multiallelic sites are plottet. A high number of multiallelic sites with low frequencies of the derived allele indicates that a high number of non-tuberculosis reads originating from environmental contamination are mapped when default parameters are used.

4 Supplemental Figure 3. Misincorporation profiles depend on the type of DNA library constructed. Sequencing data have been generated on the Illumina HiSeq2500 instrument, following two library preparation procedures applied to ancient equine subfossils (17). (a) Double-stranded library construction (11) results in an asymmetric damage pattern whereby C T damage is observed on the 5 end and G A damage is observed on the 3 end of the molecule, while (b) single-stranded library construction (3) results in a symmetric distribution of C T damage on both the 5 and 3 ends, with very little G A damage observed. The latter pattern is a more accurate representation of the true C T damage pattern on the adna molecule; the asymmetric pattern observed in the double-stranded library is an artifact of T4 polymerase exonuclease activity during end repair of 3 overhangs, and the 3 G A damage is merely the reverse complement of the 5 C T damage. mapdamage2 (6) was run on a total of 100,000 sequencing reads aligned against the horse reference genome. Red: C T mis-incoporation rate. Blue: G A misincorporation rate.

5 Supplemental Figure 4. Effects of sequencing depth and DNA damage rates on the accuracy and precision of DNA damage estimates. Helicobacter pylori sequencing data were simulated by using gargammel (14) and applying four increasing levels of postmortem DNA deamination rates (D, 2D, 5D, and 10D). The simulations used the empirical template size distribution of the H. pylori genome recovered from the Iceman (10a). The simulated data were trimmed, and overlapping paired-end reads were collapsed before being aligned against the H. pylori genome (NC_000915) using PALEOMIX (16). High-quality unique reads were downsampled (from 100,000 to 100), and cytosine deamination rates at overhangs (DeltaS) were estimated in mapdamage2 (6) using standard parameters. (a) Mean (green), median (orange), and 2.5% 97.5% quantiles of the posterior distribution of the damage parameters (whiskers). (b) Misincorporation profiles within the first and last 10 sequence positions when using 100 and 100,000 sequences simulated with the lowest (D) and highest (10D) damage levels. The C T misincorporation rate is shown in red; the G A misincorporation rate is shown in blue.

6 Supplemental Figure 5. Genuine misincorporation profiles on distantly-related reference genomes. M. leprae (left) and M. smegmatis (right) sequencing data were simulated using gargammel (14) assuming nick frequencies of 0, 0.25 as the geometric parameter for the length of overhanging ends, and 0.5 and 0.01 cytosine deamination rates in single-stranded and doublestranded DNA, respectively. A total of 1.77 million of paired-end reads were then trimmed and overlapping paired-end reads were collapsed before being aligned against three possible reference genomes using PALEOMIX (16) (M. leprae, top; M. smegmatis, middle, and; M. tuberculosis, bottom). The fraction of high-quality unique reads aligned is provided between parentheses. Misincorporation profiles were generated through mapdamage2 (6). Mis-incorporation rates of 1% are indicated with a dashed line to facilitate comparisons between the different conditions. Red: C T mis-incoporation rate. Blue: G A mis-incorporation rate.

Addressing Challenges of Ancient DNA Sequence Data Obtained with Next Generation Methods

DISSERTATION Addressing Challenges of Ancient DNA Sequence Data Obtained with Next Generation Methods submitted in fulfillment of the requirements for the degree Doctorate of natural science doctor rerum