WGS capable of revealing food fraud but limitations identified

A student from the university used untargeted shotgun sequencing to examine whether WGS could analyze foods to identify the various components of plant or animal origin as part of a thesis.

The work looked at 22 burgers (10 from Denmark and 12 from international locations) and 61 GB of raw Illumina MiSeq paired-end sequence data was generated, cleaned, trimmed and mapped against 15 reference databases using MGmapper.

Robust data analysis or macaque burger?​

Almost all the main components of the burger (patty, bun, vegetables etc) were identified correctly.

However, initial data analysis showed that between 3-5% of the DNA from each sample came from macaque monkeys. A closer look showed it was actually DNA from cattle.

The fact it was reported for all samples made it clear it was systematic error, which likely arose due to the way MGmapper assigns reads to databases – if two references with identical alignment score is found, the reference listed first is accepted as the true hit.

This shows there is a need to expand reference databases and improve sequencing technology to yield longer reads and methods of analysis.

Further investigation with nr/nt BLAST on some ~500 reads found another more likely match: cattle, according to the work.

The reads in the study were ≤ 250 bp (basepairs) long. For the macaque monkey finding 50 sequences were BLAST'ed and 303 alignments analyzed.

Building database capacity​

Rene Hendriksen, senior researcher at the National Food Institute, said the output is not better than what is inputted into the database.

“Macaque monkey aligned 65% with bovine DNA so the computer output said it was macaque monkey but a deeper analysis found cattle, it is a limitation of the database,” ​he told us.

“We have the sequencing technology up and running but matching the outcome to the database is the limitation. There are lots of salmon DNA but not much herring and mackerel, as there is no interest in this in terms of farming. ​

“It is also the case with phenotypes – cheetah aligned with bread wheat 73% and bat aligned 99%. There are limits with the database, mammals especially cause problems. ​

“If have an unexpected output do more analysis, it requires the database to be populated with more data for a perfect hit. Data will come with time and more robust analysis will be possible.” ​

Hendriksen added burgers were chosen as they are a well-known and complex commodity consumed worldwide and the number of samples was determined by it being a student project.

The work found the technology is still at a phase where application by untrained personnel is unfeasible, due to the knowledge of genomes and bioinformatics needed for interpretation of results but development of sequencing platforms, computers and algorithms could see this overcome.

Hendriksen said more research and diagnostics labs are embracing the tool.

“This will mean more data sequenced but it can be complicated to upload to repositories and this limits the data sharing and annotating DNA so we know what we have," ​he said.

“In the future it will have a role the problem is cost, who will fund it? There is consumer interest to know about a tampered commodity but not the funding, supermarkets have customer money but what would be the intention to have analysis, producers are also interested but it has to have someone who wants to know what is in a commodity or food product. ​

“Food agencies would be speculation of a problem post-event not routine surveillance of certain commodity. It is difficult to find who wants to use.”​

Bacillus cereus finding​

Bacillus cereus, which produces toxins that can cause food poisoning, was also found but this could be due to growth during transport to the laboratory.

Burgers were bought from Romania, Qatar, France, Sweden, Spain, Austria, Switzerland, Thailand, USA and Lithuania and stored for a maximum of four months before they were processed.

They were transported in the paper wrapping from the restaurants, temporarily stored at -18°C and frozen to -80°C.

Samples were not stored and transported according to proper protocol, said Hendriksen.

“We found Bacillus cereus but it was not our primary aim. It probably had been there from the beginning but we can’t say in a volume that could have caused food poisoning. ​

“It could have been a trace and been enriched through time and transportation, we just wanted to indicate Bacillus cereus was found but it may be in a small amount and cause no harm.” ​

Xanthomonas campestris, ​a pathogen of many plants and cultivated varieties of Brassica ​(mustard) was found, which could come from infected plants used by accident in ingredients of the burger.

Pseudomonas fluorescens​, often present in deteriorating milk and meat products, was also detected, which could be taken as an indicator of beginning food spoilage.