How bioinformatic and sequencing data might inform the regulatory process - OECD Bioinformatics and Microbial Pesticides Seminar - Andre Silvanovich
0 likes170 views
The document discusses the limitations of whole genome sequencing (WGS) and bioinformatics in the regulatory risk assessment of microbial pesticides, emphasizing that current assessments do not significantly benefit from WGS data. It highlights a lack of experimental evidence on protein functionality, leading to many functional assignments being mere predictions rather than conclusions drawn from data. The text argues that hypothesis-driven evaluations supported by existing experimental data may eventually enhance regulatory assessments as the understanding of protein functions improves.
How bioinformatic and sequencing data might inform the regulatory process - OECD Bioinformatics and Microbial Pesticides Seminar - Andre Silvanovich
- 1. /////////// How Bioinformatic and Sequencing Data Might Inform the Regulatory Process (Industry Perspective) OECD Expert Group BioPesticides Seminar: Bioinformatics and Regulation of Microbial Pesticides Andre Silvanovich, PhD June 24, 2019
- 2. Regulatory Risk Assessment of Microbials & Whole Genome Sequencing (WGS) The current risk assessment paradigm for microbials is: Fit for purpose Has been and is used to successfully evaluate risk for commercialized products At this time WGS coupled with genome assembly and annotation does not improve the quality or value of a regulatory risk assessment Lacking information on protein functionality
- 3. Commoditization of WGS has created unrealistic expectations for its use Evolution of Whole Genome Sequencing Rapid evolution through platform changes 454 Illumina Pacbio Quality and throughput Steadily decreasing cost per base Accessible through fee for service vendors Routinely described in popular press “$200 genome” Personalized medicine “$999 consumer-friendly genome sequence” Image and quotations downloaded from https://www.wired.com/story/whole-genome-sequencing-cost-200-dollars/ May 22, 2019
- 4. Raw data are strings of nucleotides (G, A, T, & C’s) - that are of no value without further analysis WGS Data are Meaningless Without Further Analysis 4 Assembly Combine short nucleotide strings into longer strings to construct collection of “contigs” (longer strings), or a complete chromosome Identification of single nucleotide variants (SNVs) and insertion or deletion of bases (indels) Added nucleotides that are present or nucleotides that are missing relative to a reference sequence Changes in amino acid sequence Use nucleotide sequence to identify taxonomy Specific sequence unique to genus or species reference sequence Coupled with additional experimental data (e.g., morphology and growth on selective media) Gene prediction Identify promoter-coding-3’ untranslated region (UTR) sequences, or amino acid sequence Use predicted amino acid sequence to assign predicted gene function Build an annotated genome
- 5. Sequence is not enough to inform a microbial regulatory process WGS Data has Overwhelmed our Analytical Capacity “In a study published Tuesday in PLOS Biology, researchers at Northwestern University reported that of our 20,000 protein-coding genes, about 5,400 have never been the subject of a single dedicated paper.” “Most of our other genes have been almost as badly neglected, the subjects of minor investigation at best. A tiny fraction — 2,000 of them — have hogged most of the attention, the focus of 90 percent of the scientific studies published in recent years.” Cited text viewed and image downloaded Sept. 19, 2018 https://www.nytimes.com/2018/09/18/science/why-your-dna-is-still-uncharted- territory.html?action=click&module=Discovery&pgtype=Homepage
- 6. May 2019 Release Statistics for UniProt* We Lack Experimental Data on Protein Function Downloaded from https://www.ebi.ac.uk/uniprot/TrEMBLstats May 22, 2019 Fewer than 1% of protein sequences have evidence of their existence through experimentation Greater than 99% of protein sequences are predicted from genome sequencing activities * The Universal Protein Resource (UniProt) is a comprehensive resource for protein sequence and annotation data. UniProt is a collaboration between the European Bioinformatics Institute (EMBL-EBI), the SIB Swiss Institute of Bioinformatics and the Protein Information Resource (PIR Knowledge of protein function is important to the regulatory process
- 7. Bacterial Protein Sequences Lack Data on Protein Function Thousands of bacterial genomes sequenced Millions of protein sequences A small proportion of these proteins have been studied experimentally One third do not display sufficient similarity for functional assignment Price et al., 2018. Mutant phenotypes for thousands of bacterial genes of unknown function. Nature 557:503-509 Chang et al., 2016 COMBREX-DB: an experiment centered database of protein function: knowledge, predictions and knowledge gaps. Nucleic Acids Research 44:D330-D335. Protein with experimental evidence Functional assignment based upon alignment with triangle protein Functional assignment based upon alignment with square protein Functional assignment based upon alignment with pentagon protein Triangle protein Triangle-like protein Triangle-like protein Triangle-like protein Annotation Assignment Most of the proteins have predicted function based on sequence similarity Automated pipelines create annotations (function assignments) with what is known at the time Functional assignment has not been systematically documented No standardized process for functional assignment Inaccurate or incorrect annotations are sometimes produced and propagated by pipelines A controlled lexicon is not used No understanding of the functional context is produced (Does the “gene” yield a protein) Knowledge of protein function is important to the regulatory process
- 8. Count Description 373 phosphotransferase 42 pyridoxal phosphate (plp) phosphatase 21 putative phosphatase 230 phosphatase ybha 1 conserved hypothetical protein 3 predicted hydrolase 2 ygha hydrolase 209 cof-like hydrolase family protein 3 hydrolase of the had superfamily 475 pyridoxal phosphate phosphatase 1 putative phosphatase ywpj 101 pyridoxal phosphate phosphatase ybha 2 phosphatase 2 cof hydrolase 702 pyridoxal phosphatase 1 hypothetical protein samn05216485_102158 1 hypothetical protein samn04487822_11863 515 cof-type had-iib family hydrolase 1 hydrolase_3, haloacid dehalogenase-like hydrolase 3 pyridoxal phosphatase / fructose 1,6-bisphosphatase This bacterial protein sequence is found 2689 times in GenBank NR database: MTTRVIALDLDGTLLTPKKTLLPSSIEALARAREAGYQLIIVTGRHHVAIHPFYQALALDTPAICCNGTY LYDYHAKTVLEADPMPVNKALQLIEMLNEHHIHGLMYVDDAMVYEHPTGHVIRTSNWAQTLPPEQRPTFT QVASLAETAQQVNAVWKFALTHDDLPQLQHFGKHVEHELGLECEWSWHDQVDIARGGNSKGKRLTKWVEA QGWSMENVVAFGDNFNDISMLEAAGTGVAMGNADDAVKARANIVIGDNTTDSIAQFIYSHLI Despite being the same sequence in each of the 2689 GenBank entries it is described in multiple ways: • Phosphotransferase • Phosphatase • Hydrolase • Hypothetical protein Given the count associated with each description, there are three primary annotation lineages that describe different biological activities for the same sequence Phosphotransferase is the opposite of phosphatase activity • Phosphotransferase adds phosphate, phosphatase removes phosphate What is relationship between hydrolase and phosphatase/transferase activities? Annotations of Identical Protein Sequences Vary Considerably Knowledge of protein function is important to the regulatory process
- 9. Potential Conclusions Related to Disparate Annotations for Identical Protein Sequences The protein has multiple enzyme functions and all annotations are correct The protein is a hydrolase Interpro says the following: Haloacid dehydrogenase-like family that includes cof-hydrolases, ATPases and phosphatases Haloacid dehydrogenase family that includes hydrolases and eukaryotic phosphatases Some annotations for hydrolases are due to activity versus a common lab test compound used to determine phosphatase activity There is no evidence that this phosphatase activity is physiologically relevant to hydrolase activity Perhaps the phosphorylase annotation is a book keeping error or typo that is propagated If not a multifunction enzyme, resolution of the disparate annotations would require: Investigation of the annotation lineage to determine if one or more annotations are based upon physicochemical or genetic analysis of mutants Genetic analysis of mutants Purification of the protein followed by physicochemical analysis Knowledge of protein function is important to the regulatory process
- 10. Current State of WGS Data Analysis & the Regulatory Process Tremendous strides have been made in: Sequencing platforms and technologies Software and tools used to interrogate and catalogue sequence data The gap between sequence data collection and experimentation continues to widen Given the current body of knowledge and state of the field, at this time WGS coupled with genome assembly and annotation does not improve the quality or value of a regulatory risk assessment High quality sequence data are readily collected & bioinformatic tools are robust and fit for purpose However, the total body of experimental protein data is significantly lacking Without experimental data, the majority of functional assignments are simply predictions - 1st OECD EGBP document 2001 - EPA BPPD established in 1995
- 11. Provision of Selected Microbial Sequences to Regulatory Authorities Could Add Value Hypothesis–driven evaluation of specific sequence / bioinformatic information has the ability to inform a regulatory risk assessment Examples such as: Taxonomic identification Presence of potential resistance to clinically relevant antibiotics These hypotheses are underpinned with existing experimental evidence / data that indicate they may yield meaningful hazard identification If hazard is identified via sequence / bioinformatic information, experimentation / data generation would still be needed to assess hazard & determine risk
- 12. WGS in Microbial Risk Assessment – Not Ready for Prime Time The current risk assessment paradigm that focuses experimental evaluation is: Fit for purpose Has been and is used to successfully evaluate risk for commercialized products At this time WGS coupled with genome assembly and annotation does not improve the quality or value of a risk assessment Hypothesis–driven evaluation of specific sequence / bioinformatic information does have the ability to inform a regulatory risk assessment In the future as the body of experimental protein knowledge expands assembly and annotation may prove to be of value for a hazard assessment A standardization of software tools, thresholds and references will likely be required
- 13. /////////// Thank you for the opportunity