Hi guys,
I used laser-capture-microdissection (LCM) to pick the sponge-associated filamentous bacterial cells. Then I followed the instruction of QIAGEN repli-g single cell Kit to amplify the single-cell DNA. 16S rRNA-PCR was performed to verify the bacterial taxon.
Single-cell DNA pair-end library was constructed, sequenced by MiSeq PE300 and assembled by Newbler 2.9. Lastly, it started from 3.1GB raw reads and ended up with a 2.2MB txt file, containing ~1100 contigs (>500bp).
What bothers me is that there is no 16S rRNA/23 rRNA in my assembly according to the annotation of RAST ,but before sequencing, the 16S rRNA gene could be amplified. Additionally, 5S rRNA was found in assembly.
I don't have much experience with assembling genome and this is my first time to do single-cell sequencing. I merged the reads before submitted them to newbler. Can someone suggest me a better way to assemble the genome?
Also, a collaborator is needed to help us with genome assembling and bioinformatics analysis. Credits will be shared for sure. If anyone is interested, please contact me.
Cheers,
Fang
Fang Liu - a single-cell assembly is different from common WGS fragment-assembly as the coverage of the genome in single-cell sequences are highly uneven and most multi-cell assembler assume more or less an uniform coverage.
I would highly recommend you to test SPAdes, a single-(and multi)cell assembler which produces very good results, is easy to use and runs relatively fast.
Anyhow, all assemblers have naturally problems to resolve/assemble repetitive elements of the genome at hand and the 16S rRNA operon is a repetitive region. So depending of the genome characteristics, coverage and the sequencing platform, getting the 16S assembled might be a bit difficult (scaffolding is problematic for long repeats if the insert size of the PE reads is to small).
Looks like an incomplete genome as Juan mentioned.
You can send it to companies that perform professional services for this.
Cheers
JS
Likely poor assembled genome due to poor sequence data? Please check other common used gene markers with a similar length.
Velvet or SOAP de novo are good, easy-to-use and free alternatives. I would also check the quality of the raw data e.g. using fastqc to check the Phred scores of the data.
Another advice is using a reference genome for the assembly, if this is available. You can add a reference genome in either Velvet or SOAP de novo to improve the assembly.
You can look at our genome comparison paper for details regarding genome assemby and annotation.
Fang Liu - a single-cell assembly is different from common WGS fragment-assembly as the coverage of the genome in single-cell sequences are highly uneven and most multi-cell assembler assume more or less an uniform coverage.
I would highly recommend you to test SPAdes, a single-(and multi)cell assembler which produces very good results, is easy to use and runs relatively fast.
Anyhow, all assemblers have naturally problems to resolve/assemble repetitive elements of the genome at hand and the 16S rRNA operon is a repetitive region. So depending of the genome characteristics, coverage and the sequencing platform, getting the 16S assembled might be a bit difficult (scaffolding is problematic for long repeats if the insert size of the PE reads is to small).
Fang Liu, Sebastian's suggestion to use SPAdes is very good. SPAdes is a very easy to use and fast de novo assembler, and you can even use the option --sc that optimizes the parameters assembly to single-cell data. I suggest you take a look on the SPAdes 3.5 manual, it is a very good and easy to understand starting point: http://spades.bioinf.spbau.ru/release3.5.0/manual.html
Also, you could try to play around with your raw reads file so you can improve the quality of the assembly. Try to trim them based on the quality, or reduce the coverage (to something around 50x and 20x). Most assemblers have problems assembling reads with very high coverages of your target genome. If you have a closely related genome to your species, a good thing to do would be to use QUAST (http://quast.bioinf.spbau.ru), an assembly comparison tool that can evaluate how good your assemblies are compared to themselves and to a reference genome. You could also try to map your reads to the related genome and try to assemble you genome again using only the mapped reads.
Finally, try to use the RNAmmer server (http://www.cbs.dtu.dk/services/RNAmmer/). Just upload your contigs file and RNAmmer will try to identify the rRNA sequences on it :)
Fang, if ir reads file is in the BAM format, you could use Picard's tool DownsampleSam (http://broadinstitute.github.io/picard/command-line-overview.html#DownsampleSam). Picard is a very nice java application with several handy tools for handling and analyzing BAM/SAM files. The DownsampleSam tool take a random subset of the reads, which will be a good representation of your original data. You need to calculate how many reads you need to achieve a desired coverage, and divide that number by the total number of reads in your raw file. That will give you a percentage, and you will use that as the 'P' parameter in DownsampleSam. In the end, it will give you a BAM file with the desired coverage so you can reassemble the genome.
If you need to convert your files from fastq to BAM, you could use the FastqToSam, also from Picard.
Here is the link for the Picard website: http://broadinstitute.github.io/picard/index.html
Looks like you have very high coverage (3.1Gb/2.2Mb) for your genome, have you trimmed the low quality reads and adaptor containing reads? The sequencing errors will produce low frequent kmers which may confuse the assembly software. Normlization can be a good choice, you can give khmer a try.
- Badges
- Science topic
- Similar topics
- Biological Science
- Microbiology