Hi all,
I am trying to analyse sequencing results coming from ngs on bacterial strains. They were sequenced with the Illumina method. I got sequencing results in reads and contigs. This field is totally new for me. Saying "contigs" is the same as saying "reads assembled"? So, does it mean that i don't need to assemble the reads since I have contigs? Are they already assembled?
I don't understand how I "re-build" the plasmids starting from the contigs.I mean that I want to re-construct and determine all the circular sequence of the plasmid and I don't know how to do it.. is it better to do it starting from contigs or reads?
I saw tools like plasmidSPADES that allows to assemble plasmid sequences..is this used with contigs or reads? Does it make sense to use it on contigs as input?and If not, how do I use my contigs to determine the final sequence of my entire plasmids? I am quite confused..
Is it also possible that different contigs share the same sequence? Is this the result of the method used for sequencing?
Thank you very much
Hi,
let me try to answer some of the individual questions:
> Saying "contigs" is the same as saying "reads assembled"? So, does it mean that i don't need to assemble the reads since I have contigs? Are they already assembled?
"Contig" stands for "contiguous" and "is a set of overlapping DNA segments that together represent a consensus region of DNA" -> https://en.wikipedia.org/wiki/Contig. Accordingly, it is most commonly used today to refer to the consensus sequence(s) obtained from a de-novo assembly. Hence, "contigs" == "reads assembled" could be said. Since you did not provide any further information, e.g., how were the contigs created, i.e., what preprocessing, assembler, etc., were used, it is difficult to give further advice. You could check this with the sequencing provider most likely.
> I don't understand how I "re-build" the plasmids starting from the contigs.I mean that I want to re-construct and determine all the circular sequence of the plasmid and I don't know how to do it..
Here, an open question is whether you have *only* sequences from the plasmids, e.g., because your sample was specifically enriched for plasmids? Although I am not a wet-lab person and hence do not know whether this is possible or efficient.
> is it better to do it starting from contigs or reads?
Do you have a reference genome or reference genomes? If so, you could map the reads, e.g., using bowtie2 or bwa, to those references and assess the coverage and variation, e.g., SNPs. If you do not have reference genomes, you will pretty much have to de-novo assemble the reads and hope that they can be assembled into a single contig. As you mentioned circularizing it, you might want to have a look at https://github.com/sanger-pathogens/circlator. Please note that, at least for bacterial chromosomes, circularization is rather difficult when using short reads only, e.g., only Illumina (short insert) reads but no long-range data, e.g., from PacBio.
> I saw tools like plasmidSPADES that allows to assemble plasmid sequences..is this used with contigs or reads? Does it make sense to use it on contigs as input?
plasmidSpades uses reads as input, i.e., it performs short read-based de-novo assembly.
> and If not, how do I use my contigs to determine the final sequence of my entire plasmids? I am quite confused.. Please see my above answers, e.g., Circlator.
> Is it also possible that different contigs share the same sequence? Is this the result of the method used for sequencing?
Yes and no. You might have repeats in your assembled sequences. In fact, Circlator uses overlaps/repeats at the "beginning" and "end" of the contig for ciruclarization.
Hope this helps.
Best,
Cedric
P.S. If you found this answer helpful, please consider up-voting it.
Hi Cedric,
thank you for your answer that was very useful.
can I ask what you exactly mean when you say "repeat" in the assembled sequences? You mean for example "long terminal repeats" or simply a DNA sequence present (that might belong to a gene for example) in more contigs?
Also, I noticed that if I run Blast by using a sequence of a known protein as a query against a file containing all the contigs, the known sequence matches more contigs, and most often the same sequence in different contigs can be different in a few nucleotides that result in a different identity percentage with the known sequence...why does it happen? if there are more contigs for a same sequence, should not this latter be exactly in the different contigs?
Thank you very much again
By "repeat" I mean that a certain sequence (from a few bps to, maybe, several hundred bps) may occur repeatedly (genuinely or artificially) in a genome/chromosome/plasmid. A repeat may be present in several contigs or just in a single contig. For some information about terminology commonly used in assembly, the following might be of help: http://wgs-assembler.sourceforge.net/wiki/index.php/Celera_Assembler_Terminology. The following paper might be of interest too for dealing with repeats: http://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-12-95
Regarding your protein-search, there might be several reasons why you get multiple hits in different contigs. One might be that those are spurious hits; the longer your query, the less likely that they are spurious. Another reason might be that that protein is present in multiple copies. Yet another reason might be that you have a certain amount of heterogeneity in your data, e.g., strain-level variation that leads to similar contigs with slight variation, e.g., SNPs, or even unrelated, contaminating organisms. One way to approach strain-level variation is to try to collapse the contigs, e.g., using cd-hit-est, at a certain nucleotide-identity level, e.g., 95% - 99% identity, and check whether the heterogeneity is reduced. This could potentially be peformed on the gene-level too, i.e., a gene-calling software is used to predict coding sequences from the contigs.
Unfortunately, there is no simple answer to this (e.g., should you rather work on the contig-level or on the gene-level, should the sequences be collapsed or not, should the contigs be somehow filtered at the very start, i.e., keeping only those with a "reasonable" percent- and fold-coverage) and it might be good to search for somebody with more experience in such analyses that could support you, e.g., someone from the sequencing provider and/or someone from a bioinformatic core facility. Lastly, many of these questions depend on a priori domain knowledge, e.g., was it an isolate culture, were enrichments for plasmids performed, etc.
Best,
Cedric
Thank you very much for your answers Cedric. It was really helpful. Thanks
- Badges
- Science topic