Hi guys!
I'm sequencing and assembling a de novo invertebrate genome of about 0,8Gb. I want to make a lane run of Illumina nextera mate-pairs, but I'm having some difficulty on selecting the size of the mate-pairs.
I saw that for de novo genomes is better if you have mate-pair-known sizes. So I was planning on doing 5kb and 10kb gel-size selection. However, the standard Illumina procedure is to make a random library of inserts ranging from 2kb to 15kb.
I was wondering how much bias it would insert for the de novo algoritm, lets say, allPaths, to assemble it? Does anyone of you have experience with that?
I don't know if I insist on trying to have the mate-pair with known insert size or if I just go for the random mate-pair library preparation (I have plenty of pair-ends already).
Any experinces you have, or papers to share, would help a lot! Thank you so much!
I cant answer regarding any alignment bias. I would however recommend to go for the recommended size spread for diversity reasons. The fragmentation will not be random, it will be partially sequence dependent. Hence there's a risk that you'll miss out on regions if you restrict your fragment sizes.
My guess is that it would not matter much, just be careful to estimate correctly the DNA concentration. That will determine your span of fragments
Bioinformatically it should make little difference, unless is weird genome with large repeats. My guess is that regular pair-end will be sufficient for assembly (2x250 bp sequencing of PCR free library )
One question first: Is the genome size correct? 800kbp does not sound right for me...
Regardless, based on my very limited experience with allPath and quite extensive experience with Newbler, I would strongly suggest to do a size selection. In most cases, a single 10k library will be sufficient. If you want to prepare for the worst, go for 5k and 15k. That will usually solve (almost) all problems. No size selection will also work most of the time, but the number of cases where we ran into problems with that approach was too high for my liking (~20% of all cases).
Concerning the paired-end reads I can also offer the advice to stick to the TruSeq PCR free libraries for de novo projects. Nextera will also work reasonably well for genomes with a G+C up to 65%, given that it does not contain too many repetitive regions. In contrast, using Nextera XT will most often lead to grief (and in the worst case you will not even notice the errors introduced).
HI Marcela,
I have just a quick answer for now - the best assemblies have 6+ mate pairs , the larger the insert the better but this could be difficult for beginners. I will ask around for more depth as well. Best Joe
800 Mbp makes more sense :)
I do not know how large repeats in invertebrate genomes can become, but 10k (or better 15k) should take care of most problems. If you want, I can sent you the protocol we use for library preparation.
Yes Rohan, this will be done too!
Christian, it would be great if you could send me the protocol, yes! Does it differ a lot from Illumina standard procedure?
Thanks for posting the question that is so close to my heart! In planning the future work, I have also ran into this dilemma. Specific questions as below:
1. The local rep from illumina also told us that gel-free protocol could help with the diversity issue. Does this means that more parts of the genomes can be covered for mate-pairing?
2. Previously we sequenced a 5MB bacteria genome with Nextera XT PE. We were unable to close the genome. Would this has anything to do with the use of Nextera XT, as mentioned by Christian? Would using Mate-pair increase the chances of generating a complete genome significantly?
Hey wei!
1- Yes, if you size select, you will have a more restrict library. If you sequence all the fragments, them the probability for you to have a more diverse coverage increases.
2- Using mate-pair would definitely help you to finish your genome.
For both cases, some researches have shown that Nextera enzime does not cut randomly, it rather give preferences for gene regions (probably because of GC content and specific sequences of restriction sites). So, its probably true you will have more coverage for coding sequences than for non-coding. Still, to finish a bacterial genome it should be fine. For eukaryotic genomes is that sometimes you have to have a enzime-free fragmentation library as well.
Hope I've helped! x
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools