You may try with PCR, for this, you need to design the primer, you can do this at http://www.ncbi.nlm.nih.gov/tools/primer-blast/ all you need is the sequence you want to look for, and then just determine the leng of the primers, this can help you to understand the design http://seqcore.brcf.med.umich.edu/doc/dnaseq/primers.html 

Thank you very much, I will read further. I am not sure I can predetermine the set up of the run as it is done at a center that does not mention that option but I can discuss it with them.

This paper has complete methods for amplifying a region of gag gene from MLV in human samples. It includes the primer sequences they used.

AUTHORS:   Irlbeck,D.M., Vernon,S.D., McCleary,K.K., Bateman,L., Klimas,N.G.,

Lapp,C.W., Peterson,D.L., Brown,J.R., Remlinger,K.S., Wilfret,D.A.

and Gerondelis,P.

TITLE:   No association found between the detection of either xenotropic

murine leukemia virus-related virus or polytropic murine leukemia

virus and chronic fatigue syndrome in a blinded, multi-site,

prospective study by the establishment and use of the SolveCFS

BioBank

JOURNAL:   BMC Res Notes 7 (1), 461 (2014)

PUBMED ID:   25092471

http://www.ncbi.nlm.nih.gov/pubmed/25092471

Thank you Brian. I have moved on from XMRV and realize that data obtained through classical virological methodologies is not acceptable for MLV discovery.  NGS data is going to be necessary.  I have no input in the SUNY@Buffalo NGS center and they use a standard format for NGS from outside the University.  I have to mine NGS data for the sequences that I am looking for. The MLVs I am looking for are Rauscher, Maloney, Abelson and Spleen Focus Forming Virus (SFFV).

I had an RNA NGS in 2012 which was partially analyzed at Tulane and showed increased reads for HERV-K113 and the HERV-W Env, Syncytin-1 but they haven't had time to look for exogenous MLV other than finding an increased but still not impressive level of expression of Abelson Leukemia Virus that popped up on one of their analytic programs.

I am sure you heard that Lipkin found "retroviral sequences in 85%" of ME/CFS patients and believed enough in his results to request a grant. Unfortunately the grant application scored low (I can send you that email),  He then moved on to do NGS analysis of ME/CFS bowel flora and that was funded. He then started saying he had no plans to go back to his 85% finding, saying that  85% of healthy controls also had retroviral sequences which has never been reported previously. He had not said anything about that in his original email to me.

I am most interested in SFFV because of unpublished work from the NIH/NCI by an investigator who you may know who doesn't want to be quoted in a publication.  Here is what he said: 

"Hi Michael, we did a transcriptome analysis of human prostate tumors using an "pilot" array that was designed to detect viral sequences. We made the observation (about 2009) that a subset of prostate tumors contained SFFV-like sequences. It indicated the presence of a SFFV-like virus in human prostate tumors. We did not pursue the observation any further. With the availability of TCGA data for many human prostate tumors, one could mine these deep sequencing data for presence of SFFV sequences. Those data would be available in the "unannotated" sequence reads for the tumors.

We recently published a paper on HERV-K re-activation in prostate cancer as an early disease detection marker in Carcinogenesis. If SFFV is presence, one should find those sequences.

Is your antibody good for IHC?

Best, S."

I don't think anyone will do what Dr. S.A. suggests can be done with NGS. I am trying to get myself up to speed so that I can look for retrovirus in my NGS and others and go forward with what Dr. S.A. suggests. I have no training in genomics so it may be unrealistic to think that I can, but I want to try.

It's good to hear from you,

Michael

It sound like you already have sequence data and now you want to search through those sequences for retroviral sequences.  It is relatively easy to convert the NGS sequence data into a BLAST-formatted database, and then use the BLAST search program to search this database with each virus you are interested in, such as MLV.  However, this will miss any sequences of unknown viruses.   All NGS data has some contaminating sequences as well as the genuine sequences from the sample, due to contamination of the reagents used in NGS, but you can sort the wheat from the chaff after you find the sequences you are interested in.

A second, more difficult, approach would be to sort all the NGS sequences from your sample(s) into sequences which are found in the human genome and sequences which are not found in the human genome.  The "not found" sequences will include any viruses, previously discovered and sequenced or not, bacteria which were in the sample, and the NGS contaminant sequences. 

http://www.ncbi.nlm.nih.gov/bioproject/PRJNA28335

Is the bioproject for James Watson NGS genome, for an example.  This project was sorted into human and non-human sets, plus a mitochondrial genome contig.  I understand a bit about how this is done, but I am no expert and would not be able to help you find the tools needed to do this type of thing.  You would need help from a human genomics group.  The BioStars site would be a good place to start.  See for example:  https://www.biostars.org/p/51577/   Identifying Viral Sequences In Next Generation Sequencing Data

Thank you for the good advice Brian. Yes, I did my RNA NGS in 2012. I am looking for known murine leukemia viruses that have known sequence data. If I find anything I would have NGS done on a few of my cancer patients who have the same immunological footprints suggesting retroviral infection.  The RNA NGS done in 2012 cost me $1,200 and I would be willing to invest some of my own money to go forward if my results are positive..

Stay in touch,

Michael 

Hi Michael,

I've been busy spotting viral sequences in tumor RNASeq recently.

There are lots of metagenomics packages to try such as Pathseq and SURPI:

http://www.broadinstitute.org/software/pathseq/

http://chiulab.ucsf.edu/surpi/

Pretty much, you put in sequence data in bam, fastq or fasta format and the tools find the likely viral sequences for you. Whether the predictions are true or not is another matter ...

I've been really impressed by Kraken it was, easy to set up and run and is super fast and it contains murine leukemia virus sequences in RefGene.

http://www.biomedcentral.com/content/pdf/gb-2014-15-3-r46.pdf

You probably need a computer person to run it for you if you're not unix-based.

Hi Michael, Its a straight forward process to blast your sequences against MLV genomes or even all gammaretroviruses for that matter. The problem you will run into is that you will get a lot of hits for short sequences because the human genome is full of retroviral junk and some short sequences in the human genome are an exact match to the MLV genome (say 10 to 20 nts). Also MLV sequence contamination from PCR reagents is a genuine problem too. Once you identify a contig with sequence homology you will need to blast the entire contig against the human genome to make sure its not endogenous. Ive done a lot of this as you may imagine and I have a guy that does my analysis and he has the ability to do what you want so if you really want to do this I can give you his contact info. Cheers

Thank you Scott and thank you Vince, I will read up on your advice.

I have attached HERV analysis of my 2012 RNA NGS. As far as I can find, this is the first demonstration that there is simultaneous overexpression of two unrelated HERV families in one individual.  I haven't tried to publish this yet because I want to repeat the NGS and see if the expression decreases with 2 extra years of ART.  My disease processes are chronic lymphocytic leukemia and Cancer-Related Fatigue AKA fibromyalgia/ME/CFS and they do appear to be responding to ART.

The HERV-K113 analysis also includes reads for non-human viruses, Shamonda and Simbu which is unexplained but perhaps represents errors in the initial NGS steps. Note there are a fair number of reads for the Abelson MLV and only one read for Moloney MLV and none for MLV-VP62 (XMRV) so there does appear some selectivity in the NGS results.

Michael Snyderman

I have two questions. First, Pathseq and SURPI: appear to subtract the human genome from the NGS to allow identification of none-self sequences. As HERV-H and HERV-W have MLV-like sequences, won't that make it impossible to identify and exogenous MLV? Second,  Kraken appears to require an exact fit which is a problem with retrovirus because of poor fidelity of replication and therefore some sequence variation.

Are these questions a real concern?

Michael Snyderman

Hi Michael,

I didn't initially realize that you were looking for endogenous retrovirus that had mobilized. Maybe subtracting the reference genome wouldn't be a good idea so Pathseq and SURPI are out. Kraken does a neat taxonimic matching based on exact k-mer matches rather than matching the whole read. And it doesn't care about the human reference sequence unless you want it to. For example, it tells me that I have 1000 k-mers matching any papillomavirus, 900 match the human papillaoma virus, 800 match HPV 18, for example. So even if the exact sequence that you're looking for isn't in the database you'll have some idea what you've found. It will also dump all of the matching reads to a file for you to do a de novo assembly or something else fancy.

Thank you Scott. The retroviruses that I am looking for were ERVs in mice, were replication competent and jumped into our species most notably with the live virus vaccines that were attenuated in mice, the Yellow Fever and the Sabin and Koprowski oral polio vaccines.  The inventor of MMR, Maurice Hilleman as per his interview available on youtube.com/watch?v=H77ktqya8MU mentions it in passing, the interview was mostly about SV40.  The association of the virus with cancer has been documented a number of times but is still not accepted. 

Looking for this retrovirus is my passion but I am desperately trying to come up to speed so I can go forward.   The different strains of murine leukemia virus have known sequences so I would have something to match to but being retroviruses, there is always some sequence variation. If Kraken can tolerate some variability then I will concentrate on learning the program.

My son works with Linux in designing computer software for the pharmaceutical industry and is looking at Kraken now and could set me up to run it. Besides looking into my RNA NGS and those of a few colleagues with cancer, I would like to look at tumor genomes that have been posted. 

Michael

Scott, how do you decide what size k-mer to use or does Kraken do that automatically?

There is a direct application for your finding of HPV, there is an oral derivative of cidofovir, brincidofovir that is in development by Chimerix for I believe JC viral infection in transplantation. Brincidofovir should treat all DNA viruses including HPV and EBV. I am told that Chimerix will entertain looking at other indications later this year. It should be effective against HPV associated cancer. Using antivirals against viral associated cancer works and is in the literature. I can send you references.

Michael

Hi Michael,

We've done a fair bit of this kind of work fora  variety of reasons - as others have said if you aren't up to speed with linux/writing your own scripts you will probably need a bio-informatics type person to help you do this. For non-programmers who want to have a go our bio-informatiions recommend this:  https://usegalaxy.org/ as you can "pick and choose" modules for your "pipeline"and set it to run on a webbased server. Its relatively intuitive for non-programmers. Of the commercial software we've used CLC-genomics workbench (though its not cheap to buy) for students etc and it is also fairly easy to use (though the size of the files you are using generally makes it impractical to run this off a standard desktop computer). I take it from the commentary above you are looking for MuLV type sequences in human data? We did a similar exercise for novel ALV loci in different breeds of chickens cw the reference genome (for an unfunded project) . The basic analysis pathway went something like this:

• Map the reads to the jungle fowl genome reference sequence (which indicates those regions that are missing when compared with the reference genome)

• Take unmapped reads and assemble contigs using de novo assembly (which picks up those regions added when compared with the reference genome).

• Run our ERV prediction pipeline on both contigs and reference chromosomes to identify ERV loci

• Phylogenetic analysis and comparison with existing avian ERV loci databases (eg Repbase) to annotate novel gains/losses

What you will need is a pipeline for identifying retroviral sequences  in contigs/chromosomes. We use an "inhouse" method based on exonerate, others use blast based strategies or the "retrotector" suite (all have their plusses and minuses) 

Re sequence similarity between MuLV and HERVs, yes it could be a problem but if you set your mapping software to a high stringency, they are generally different enough to distinguish between them 

To get round the PCR contamination with MuLV sequences (which is a real problem with methods this senstive) you would be better off looking at integrated proviruses (DNA) rather than RNA data as if you can map the virus and flanking sequence you can be more sure that it is in the new host genome and not just a contaminant, 

The biggest problem is not fishing the data out of the NGS dataset but sorting and interpreting it all afterwards. 

Good luck and if you need any help drop us a line (like I said we do this sort of thing a fair bit) 

Thank you Rachel, I appreciate the leads. 

Michael