"Bacterial Pathogen Genomics at NCBI" presentation at the Standards for Pathogen Identification via NGS (SPIN) workshop hosted by National Institute for Standards and Technology October 2014 by Dr. Bill Klimke.
3. FDA, USDA, CDC State, Local and
Foreign Public Health Agencies
Industry/Academia Additional
DATA ANALYSIS
DATA ASSEMBLY AND
STORAGE and Analysis
DATA ACQUISITION
NCBI, EMBL DDBJ (INDIS)
(Public Access Database)
Our Current Model – Publicly available data
National Network of SequencersIntrenational Network of Sequencers
4. Automated Bacterial Assembly
SRA Reads
sample 1
Trim reads
(Ns, adaptor)
Reference
Distance tree
Find closest reference genome(s)
ArgoCA (Combined Assembly)
De novo assembly panel
Argo (Reference
assisted
assembly)
SOAP denovo
GS-assembler
(newbler)
MaSuRCA
Celera
Assembler
Reads remapped to combined assembly
Contig fasta
Read placements (bam)
Quality profile
SPAdes
5. WGS & Epidemiologically Relevant Distance (ERD)
• WGS allows high resolution genotypic comparison of
pathogen isolates
• What is the epidemiological relevance of genotypic
distance?
• Many methods to compute – we need some common
principles…
6. Since all approaches start with sequence reads, we must
retain for independent confirmation
0
0.2
0.4
0.6
0.8
1
0 500 1000 1500
Millions
FDA-CFSAN: microbial foodborne pathogen
research
SRA format bytes per sequenced base versus
number of bases in MiSeq runs
With Quality Without Qualities
0
0.2
0.4
0.6
0.8
0 200 400 600 800 1000 1200
Millions
OXFORD University: Population Genomics of
Mycobacterium tuberculosis
SRA format bytes per sequenced base versus
number of bases in MiSeq and HiSeq runs
With Quality Without Quality
Storage is manageable…
7. Reliable, transparent, high throughput, high
resolution ERDs?
Major challenge is to distinguish independent
events (SNPs) from single events that generate
multiple nucleotide differences
i.e. collapsed repeats and other artifacts,
alignment errors (reference-based alignments),
sequence quality, & recombination
11. Table: Samples currently processed (as of Sept 5, 2014) in NCBI Pathogen Pipeline
Organisms
Center Listeria Salmonella E. coli Total
CDC 903 903
FDA + State Partners* 858 6129 307 7294
100K 565 34 599
FERA 14 14
Total 1775 6694 341 8810
Processing Status
12. How to measure the system?
need the raw data (sequence reads) in unprocessed form
any read trimming/filtering along with the assembly can be regenerated
13. Assembly metrics
map the reads back to the assembly and generate a profile of each position
(coverage, alleles, qualities)
compare the assembly against other assemblies of the same organism (genus,
species) and check the expected genome size, or similarity to related genomes
annotation metrics such as frameshifted proteins
14. What is the actual measurement for sequence
similarity?
the number of pairwise SNPs between two genomes
What is the threshold?
a pairwise distance (an observationally determined cutoff below which a cluster of 2
or more isolates are considered significantly close enough to warrant further investigation)
15. Sensitivity vs. Specificity
sequence clustering
sensitivity – measure of isolates which belong to the cluster within epidemiologically
relevant distance
(true positives) / true positives + false negatives (not correctly identified)
specificity – measure of isolates which are excluded from a cluster within
epidemiologically relevant distance
(true negatives) / true negatives + false positives
29. Assembly for sample SAMN02727350
Type
Number of
contigs
Sum of contig
lengths
Full assembly 667 5251272
contigs with Listeria hits 37 3031650
contigs with Staphylococcus
hits 630 2203573
34. Table: Assembly stats for SAMN02693748
measurement result
num_input_reads 4212706
aligned_reads 4040070
assembly_num_bases 3180478
assembly_num_contigs 50
assembly_N50 2817733
poor_quality_support_bases 132321
35.
36.
37.
38.
39.
40. Organism Biosample SRA Run Similarity to:
Listeria monocytogenes IEH-NGS-LIS-00100 SAMN02567873 SRR1207486 Listeria SLCC7179
SRR1220750 Listeria J0161
Salmonella enterica Enteritidis MDH-2014-
00798 SAMN02741943 SRR1553852
Schwarzengrund str.
CVM19633
SRR1272871 Enteritidis str. P125109
Salmonella enterica Fluntern MDH-2013-
00153 SAMN02378158 SRR1067624
Javiana and
Schwarzengrund
SRR1395304 Cubana and Agona
41.
42. Proficiency Testing
• Replicate results (phylogeny, SNPs) from published studies
• Resequencing
same isolate on multiple platforms
same isolate in multiple libraries
same isolate in multiple labs
• Blinded submissions
already-characterized isolates
mixed sample isolates
metagenomic isolates
• Corner cases
Extreme coverage
Duplicates
Sample mixups
43.
44.
45.
46.
47. Acknowledgements
National Center for Biotechnology Information – National Library of Medicine – Bethesda MD 20892 USA
Richa Agarwala
Azat Badretdin
Slava Brover
Joshua Cherry
Vyacheslav Chetvernin
Robert Cohen
Michael DiCuccio
Mike Feldgarden
Dan Haft
William Klimke
Arjun Prasad
Edward Rice
Kirill Rotmistrovskyy
Stephen Sherry
Sergey Shiryev
Martin Shumway
Tatiana Tatusova
Igor Tolstoy
Chunlin Xiao
Leonid Zaslavsky
Alexander Zasypkin
Alejandro A. Schaffer
Lukas Wagner
Aleksandr Morgulis
David Lipman
James Ostell
NCBI
• This research was supported by the Intramural
Research Program of the NIH, National Library of
Medicine. http://www.ncbi.nlm.nih.gov
CDC
FDA/CFSAN
NIHGRI
UC-Davis
USDA
Vendors: PacBio, Illumina, Roche