Problem about Genometyping for sequencing base on IIlumina sequencing

[ducanh113]

i want to ask about role of barcode in barcode adapter (or in llumina sequecing) in GBS (genometyping by sequencing,ligate by overhang of Restriction enzyme,not 3'A overhang same llumina)
now,i want to preparing libarary contruction,can you speak step for prepar it?thanks for your support!

[SNPsaurus]

The barcode is a short sequence that identifies the sample. The idea with GbS methods is to sequence many samples in a sequencing lane or run, and so there needs to be a way to determine which sample a read belongs to. If 96 samples were prepared with 96 different adapters, then all the samples can be sequenced in the same lane, and later the sequence reads can be grouped by the barcode sequence. Running multiple samples in a lane is called multiplexing.

The protocol for GBS can be found at the Cornell website:http://www.igd.cornell.edu/docs/buck...l_20110808.pdf

If you are using one of the related methods you can usually find a protocol associated with the paper describing it or online. What is your project?

[ducanh113]

thanks you so much!
a about my project base on bulk segeration analysis, and i want to find marker SNP polymorphics between bulk resistance and susceptible, link to gene resistance...
about choose REs for digest, why do you you use ApeKI (rare cutter follow protocol of Cornell, and do not use BamHI or PstI or....,and mono digset or double digest,what is different?i think when use double digest, have very much small fragment for sequencing?
thanks for your support!

[SNPsaurus]

Let me answer your question by giving a short history of the different methods.

RAD-Seq (Johnson lab)
RAD-Seq was one of the first genotyping by sequencing methods for Illumina. It uses a rare-cutting enzyme (SbfI or PstI) to select loci for sequencing. The method uses shearing to convert the long restriction fragments into "tags" short enough to be sequenced on Illumina. Some criticisms are the many steps of the protocol and that cut sites near each other in the genome don't efficiently produce tags, so some missing data is possible. The typical RAD library will sample 20,000 loci at high read depth and get high quality data at each locus.

GBS (Buckler lab)
GBS was an attempt to create a streamlined protocol, enabling the rapid processing of many samples. It cuts with a frequent-cutter like ApeKI, creating fragments that have the restriction sites at both ends, then sequences just a portion of these at low read depth. While this creates many missing datapoints, for the typical application of genotyping RILs the missing data can be imputed and the parental alleles mapped along the chromosome efficiently.

ddRAD (Hoekstra lab)
ddRAD was a merging of RAD-Seq with GBS, combining a rare cutter like PstI with a frequent cutter to create short fragments without shearing. The fixed length of the tags means selecting different size ranges will select different numbers of loci, giving flexibility for the number of markers desired. On the other hand, any variation in size selection means markers will be sequenced in some libraries and go missing in others. The use of a frequent cutter means that polymorphisms are likely to create new cut sites within a DNA fragment, so missing data can be a significant problem with higher polymorphism rates.

So for a bulk segregant analysis, you don't need many markers since your resolution will be limited by the recombination breakpoints produced in the population. But you do want to accurately assess the allele frequency so you can identify the region of linkage to resistance. I'd use SbfI unless your polymorphism rate is very low, and either RAD-Seq or ddRAD. Either would probably work well. Since you can prep all your samples in one library, you don't have to worry about the size selection variation problem in ddRAD.

[ducanh113]

yes,thanks you so much
Now with Mungbean,i am waiting for refenrence complete genome for choosing RE for RAD-tag,GBS or ddRAD,i fell to worry about fragment restriction
i will calculator in silico about number of fragment restriction?how do you think?
what should i choose method for BSA (RAD-Tag,GBS and ddRAD)?
i do not identify the region linkage to resistance,try to use SSR marker but i don't have much SSR to do this (complete genome mungbean not really)

[SNPsaurus]

I wouldn't worry too much about doing in silico digests. Since you work with a plant, you'll want to use a restriction enzyme blocked by CNG methylation found in repetitive regions of plant genomes. Mung bean is a pretty small plant genome, so you could use SbfI and sequence ~10,000 tags or PstI and sequence ~40,000 tags. Since it is a bulk analysis, you probably have 2 pools- resistant and susceptible. Sequence each pool to 100X depth (SbfI 1M reads each or PstI 4M reads each). A MiSeq would work for that. Use multiple barcodes per pool so the first nucleotides have full diversity, and add in some shotgun genomic with a different barcode to help with the diversity in the restriction site region. Either RAD or ddRAD would do this just fine.

[ducanh113]

now,i don't know this mean "CNG methylation", it can "CpG methylation"?, use enzyme block by methylation in repetitive regions?
now,i have 2 bulk, bulk resistance and bulk susceptible, probably!about 20 samples or more, how much number of barcodes need to use?
how can i prep libarary of barcodes?
I don't understand when "use multiple barcodes per pool so the first nucleotides have full diversity"?
and if use SbfI and PstI,it probarly enough for identify SNPs polymophics between 2 bulks?
now,have new paper publish a marker SCAR link to resistance gen (but so far 6,8 cM) but not mapping loci,can i use it for GBS,RAP?

[ducanh113]

i don't know more about NGS,i am trying to improve it
thanks you for your support,i hope my project will have good result!

[SNPsaurus]

Plants methylate CNG (any nucleotide allowed in the middle position). The repetitive regions are often heavily methylated. Some enyzmes (like SbfI or PstI) will not cut methylated sites, which is good because you don't want to be sequencing those regions. If you have 2 bulks, then you need at least 2 barcodes. But Illumina basecalling gets confused if every cluster has the same nucleotide incorporated during sequencing, so if you barcode the resistant bulk with TGTCAG and the susceptible bulk with TGAGTC it will have problems with the first T and G since they are the same in both bulks. So use several barcodes in each bulk (2 each, for 4 total, that give a CATG at each position).

I'd use PstI, and get lots of new markers all linked much closer than the SCAR marker.

[ducanh113]

oh,thanks you
SCAR devoloped from RAPD marker,size of RAPD about 2000bp
if compete genome of Mungbean really, I want to scan on complete genome, located this marker and located location of gen resistance (single gen,not QTLs) and then focus develope in loci linkage resistance,how do you think about it?
and recently study about GBS of soybean,use ApeKI for digest after analysis in silico,i think genome Mung bean and soybean can similarity?
thanks you!

[SNPsaurus]

You could scan the reference sequence for the SCAR sequence, but since it is so far away it won't help you that much in finding the gene. RAD or ddRAD will identify 10-100 linked markers. Those can then be used to scan many samples to find the few markers always linked, and those will be in a very small region around the resistance mutations. Using ApeKI will cut the genome into 500,000 sites, so that would need many reads to get to a high depth of coverage. Better to use PstI, and a method designed for high read depth like RAD or ddRAD.

[kmcarr]

Problems when restriction site is "in phase"

Our core has been asked to sequence some ddRAD libraries prepared by the submitting lab. They have basically followed the Hoekstra protocol but came up with their own set of 192, 6bp inline barcodes. Obviously plenty of diversity in the barcode (first 6 cycles of read 1) but I am concerned about the fact that the restriction site (I believe they used SbfI as the rare cutter) is now in phase, meaning there will be absolutely zero base diversity in cycles 7-12 for their libraries. I know Illumina has made great strides in mitigating the low diversity problems but I'm not sure if this library design will sequence well without some significant spike in of PhiX.

Have you folks doing ddRAD on a regular basis worked with libraries where the restriction site is in phase? If so what, if anything do you do to mitigate the diversity issue?

Thanks.

[SNPsaurus]

RAD-Seq has similar issues. One easy way to get around it is to use barcodes with a variable spacer region of 0-3 nts. But it sounds like the libraries have been made already! We usually have some shotgun library that could use a few more reads, so we do spike those in at 30% or so if the barcodes don't have a spacer. I would also suggest not doing the usual 'slightly overload the clusters to maximize reads' approach, as low complexity and high cluster density don't do well together.

One of the things I like about nextRAD is that the sequencing primer matches the selective sequence so the whole read is high-complexity and can be used for SNP finding. Even with precautions some runs do fail with restriction-based methods, and for a facility it can be a problem trying to figure out who gets the blame for that.

[ducanh113]

Hi everybody
Now, i'm preparing libarary for GBS, i choose ApekI follow Bulker lab's protocol, and which program i should be use for analysing and calling SNPs (single end)? maybe, can i use STACK pipeline without reference genome of Mungbean?
thank you!

[rosatoc]

I do not do the data analysis, but the primary pipeline is TASSLE (http://www.maizegenetics.net/tassel/...ipelineGBS.pdf) and in the Intro section there is a suggestion to use UNEAK if there is no reference genome. I hope this helps. Good luck!