I have received two NGS files from an NGS company, both are FASTQ files that correspond to reads from a tumor sample. I heavily suspect that MCPyV is present, and I am hoping to identify it.

What downloadable software can I use for viral-host integration detection? I have access to both Windows and Mac OS, but have no experience with Linux.


2 Answers 2


You could start by trying out a metagenomics analysis program like kraken2. There are a few databases that combine bacteria, fungi, and viral searches.

One benefit of using kraken2 is that it has options to output the most likely taxa of matches at the sub-read level, which can help with identifying integrated or chimeric sequences.

I've written up the higher-level approach in more detail in this answer:

How do I measure the proportion of different bacteria in a sample from a high-throughput sequencer?

The sequence-level hypotheses about kmer origin can be found in the direct output of kraken2 (last column):

$ kraken2 --threads 10 --db /media/minion/minion_temp/db/kraken/metagenome --memory-mapping fastq_pass/barcode0?/PAK83314_pass_barcode0?_*.fastq.gz --report kraken2.report.txt > kraken2.out.txt
$ head kraken2.out.txt
C       a3e0b7c9-e6f3-40bf-900c-b1c59eb4977d    9606    208     0:70 623:3 0:53 9606:1 0:6 9606:32 131567:5 2759:4
C       8030d33f-3838-4ed3-bab6-5f3b79fd059e    575614  164     0:68 623:1 0:46 575614:2 0:13
C       5d87a648-799d-451f-b747-32a0fdf2517a    9606    190     0:108 9606:5 0:43
C       8aba667d-4a5b-4f8b-9d18-1b5f761b87b2    9606    267     0:124 9606:9 0:99 1:1
C       1b3501c2-6766-4ee8-b984-9e92470f37da    9606    335     0:70 28901:5 0:12 35885:1 0:23 9606:16 0:83 9606:17 0:6 9606:2 0:9 9606:5 0:11 9606:41
U       75c2625d-ab87-4614-8d0d-c9d12c7d1e19    0       242     0:208
U       902b0ec1-80d9-4e07-80c8-29a79191d931    0       171     0:137
C       56b7c8a8-4cfd-44ec-8922-62871b29d31c    9606    338     0:103 9606:5 0:22 9606:5 0:1 9606:13 0:55 9606:27 0:38 9606:3 0:32
C       68bf6718-2692-46e7-ba2d-f45952fd6138    9606    197     0:68 1:9 0:38 9606:19 0:29
C       81f0f4ae-3bb5-4b94-89ff-152ff3b6f5d7    9606    238     0:146 9606:46 0:12

I'm aware that you're interested in looking for integration of a single viral genome. This can be specifically looked at with kraken2 because it can handle custom-built indexes as well. However, the benefit of using a comprehensive index (e.g. the MinusB database from here) is that it will make it much easier to exclude the other numerous endogenous viruses that are already integrated into the genome, whereas using a MCPyV-only index may lead to false-positive matches to sequences that are from another virus.


FYI (non-OPs) it is very likely this sample has already been diagnosed and proven positive via a PCR test. I don't know the OP, but I am pretty familiar with oncovirus studies. To be clear, my guess is that PCR positive - doesn't mean cancer.

  1. What I think is occurring (if I'm wrong hopefully the OP will say) - is the OP has a cancer patient(s) and has good evidence that the underlying cause is MCPyV: which is a very specific oncovirus.
  2. What they want to do is identify viral integration in the patient genome as evidence of oncogenesis. To put this general field in perspective I think this issue is literally the rise and fall of many investigators and companies in cancer (again not focusing on a singular virus per se but generically on the oncoviruses). Keep in mind I have an explanatory bias towards viruses.
  3. The molecular genetics of integration is likely to be complex.

In summary, the OP will particularly want to do (I suspect) is map the viral integration site. What is difficult is not diagnosis of the virus because at any time post-any sort of approximate assembly the genome can be locally blasted against the reference virus and its likely to have already been PCR'd. It's getting an assembly good enough to understanding the integration site - to demonstrate integration has occurred and thats is likely difficult (but I hope I'm wrong). Inverse PCR was used for this sort of thing - but thats a wetlab approach and PCR-based strategies with traditional Sanger sequencing can still be very useful here, but its laborious but was/is a traditional approach.

Thus, I suspect the OP is from a classical virological background venturing into NGS. I noted the OP used a human reference genome hg38 in a previous post.

If thats correct then ... the big issue in my opinion is obtaining a correct assembly of the putative virus-human genome O Steps would be,

  • QC
  • Assembly e.g. via bwa-mem2 ... O O
  • Construct a local makeblastdb database of the genome then ..
  • blastn between the human genome and a viral query to identify the integration site
  • Perform something like a gatk analysis to have a different perspective on the integration site via VCF
  • hone in on the contig via IGV for optical analysis and study how the virus has integrated

O My eukaryotic genomics is outdated and the above approach is summarised in Genomics in the Cloud by Geraldine A. Van der Auwera, Brian D. O'Connor published through O'Reilly Media. I just focus on small genomes.

O O and NOTE I cannot remember if bwa-mem2 uses a reference genome e.g. hg38 type thing (Auwera and O'Connor will describe it). If it does the assembly of the integration site, in my opinion, would be compromised and alternative assembly strategies might (probably would) be required. I don't think this will be trivial - I think it would need a skilled eukaryotic bioinformatician who could be told about the biology of viral integration. Keep in mind, eukaryotic genome assembly is not my thing.

However, riches await the investigator if this can be satisfactorily solved IMO.

To answer @gringers point, Kraken2 would be useful for viral integration per se and yeah the results would be interesting because there's quite of them ... but in this case - I think - its an exact genetic sequence of a single species, thus I think direct blast-type blasting or minimap2 mapping is a better strategy than Kraken2. Just my opinion.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.