The role of Metagenome sequencing in the discovery of COVID-19 (Section Two)

In terms of bioinformatics, recent studies have evaluated the pros and cons of different methods of bioinformatics analysis. Not only the existing methods have different algorithms, but also the databases used are different, resulting in differences in pathogen identification capabilities and relative abundance calculations. In the absence of a unified standard for the mNGS biosynthesis analysis process, users' choice of biosynthesis analysis software based on personal experience, accessibility, and simplicity will affect the repeatability of the experiment and the reliability of the results, becoming an obstacle to clinical standardization of mNGS.

Therefore, this research focuses on two important links of nucleic acid extraction and bioinformatics analysis, and compares and evaluates three human host removal kits (Ultra-Deep Microbiome Prep, QIAamp DNA Microbiome Kit, Micro-DXTM) and a variety of existing commercial Pros and cons of commercial bioinformatics tools.

The researchers collected samples of 9 individual fluids (peritoneal fluid, pus fluid, joint fluid, sputum) and a bone tissue specimen for sequencing, and then used different bioinformatics software such as Unix-based systems (Kraken, Metaphlan2, and MIDAS), based on Web version of commercial or non-commercial (BaseSpace, Taxonomer and CosmosID) analysis software and commercial workstations (CLC genomics Workbench). They found that the amount of offline data and the proportion of human-derived sequences in different samples were significantly different (3.5% -98.9%). The human-derived proportion was independent of the sample type, the characteristics of each sample and the efficiency of the dehosting kit related.

Figure: The mNGS biometric analysis software involved in this article

Using culture or MALDI-TOF as the gold standard, researchers evaluated and calculated the number of pathogens identified by each analysis software and their corresponding positive, false positive, and sensitivity parameters. In clinical practice, new technologies require both high sensitivity and high positive predictive value (PPV). Using two popular software, Kraken and Taxonomer, you can see that it has detected tens to hundreds of microorganisms, and the calculated PPV is extremely low. Although setting threshold filtering is helpful to solve this problem, there is still debate on how much threshold is set. In addition, the software needs to comprehensively use multiple parameters, such as relative abundance, genome size, genome coverage, and other auxiliary pathogen identification.

Researchers also detected 1% of human P. bacillus in negative quality control samples. Their analysis of P. bacillus may be caused by environmental or sample contamination, introduction of sequencing reagents, or fuzzy comparison of biological information analysis. The problem of background bacteria has also been mentioned in several studies. Do these detected pathogens actually exist in the samples? How to distinguish between pollution, colonization or disease? Technologists, bioinformatics staff, microbiologists, and clinicians also face escalating challenges. In summary, the current clinical application of mNGS faces four challenges:

1. How to reduce the socioeconomic cost as much as possible

2.How to determine the correlation between multiple pathogens detected by metagenomic technology and clinical infection

1. How to improve the timeliness of detection and achieve real realtime?


2. How to detect the reported pathogens (new pathogens) that are not collected in the database?

Faced with the above challenges, we also encountered a series of questions. The editor has collected two questions often asked by mNGS clinical experts. From the perspective of current technology development, the following tentative answers are given.

Question one: Why are some samples culture positive, nucleic acid test positive, but mNGS did not detect out? Doesn't this technology work?

Any clinical test must consider the economic cost of health, which is also one of the considerations of the new technology mNGS, so that there is a trade-off between the current economic cost of mNGS and sensitivity. Clinical samples have a fairly high degree of complexity, and many pathogens are low in content after infection or sampling after treatment leads to a reduction in the number of pathogens. In the case of limited data volumes, target pathogens are lost due to too little information.

Because of the detection limitations brought about by these challenges, it often leads to the clinician's recognition limitations, so we hear the clinician's disapproval of the function of mNGS. Although expanding the volume of data is one of the means, it is always limited within the limits of health economic costs. Enriching pathogenic bacteria, increasing the amount of sequencing data, and thereby improving sensitivity; detecting long fragments, and then improving specificity are the current and future directions of our joint efforts with the clinic.

Question two: Clinicians often report that the mNGS test result list many pathogens. Which pathogen or pathogens are the culprits?

The detection of multiple pathogens is mainly caused by the following two aspects:

The mNGS algorithm has short sequence fuzzy alignment, especially for some species with higher homology, so its classification significance is discounted in such species with higher homology. How to solve the problem that a short sequence compares to multiple pathogens at the same time? At present, our R & D personnel are rapidly developing characteristic gene databases. The use of multiple databases will further improve the accurate comparison of pathogenic bacteria.

Multi-pathogen co-infection, multi-pathogen infection of primary and secondary infections or respiratory secretions, and the microbial diversity of the open system of intestinal secretions will lead to detection of multiple pathogens. This is not actually a limitation of the technology itself, but a clinical medical problem. The culture method, mass spectrometry method, and multiplex PCR method can also detect multiple pathogens. The final diagnosis conclusion requires the intervention of relevant information of clinical diagnosis and the joint participation of clinical infection doctors. It is not possible to solve all problems by testing alone. In the case of sufficiently streamlined reporting, the advantage of multi-pathogen list display is to provide possible effective clues on the one hand, and to assist clinical diagnosis when accurate and clear clinical information is available on the other hand.

For mNGS, we can say that a new window has been opened for etiological diagnosis.

Further reading: Introduction to Shotgun Metagenomics, from Sampling to Data Analysis and Applications of Metagenomics in Biotechnology and Health Care.

Reference

1. Couto, N., Schuele, L., Raangs, E. C., Machado, M. P., Mendes, C. I., Jesus, T. F.& Autenrieth, I. B. (2018). Critical steps in clinical shotgun metagenomics for the concomitant detection and typing of microbial pathogens. Scientific reports, 8(1), 13767.


2. "Genomic epidemiology of novel coronavirus (nCoV)". nextstrain.org. Retrieved 11 February 2020.


3. Gyarmati, P. et al. Metagenomic analysis of bloodstream infections in patients with acute leukemia and therapy-induced


4. "WHO | Novel Coronavirus – China". WHO. Archived from the original on 23 January 2020. Retrieved 1 February 2020.


5. P Sci. Rep. 6, 23532 (2016).


6. Hasman, H. et al. Rapid whole-genome sequencing for detection and characterization of microorganisms directly from clinical


7. Chu JT, Hossain R, Silverblatt FJ, Hyle EP, Turbett SE. Case 22-2017. A 21-Year-Old Woman with Fever, Headache, and Myalgias. N Engl J Med. 2017 Jul 20;377(3):268-278.