Hello Qinhong,

What is the average quality of your reads? Can you post the fastqc plots? Perhaps your sequencing run wasn't good at all.

You need to check all the parameters of your trimming to ensure you're only using high quality reads. Can you let me know what were the trimming option you used?

All the best,

Unfortunately a bad sequencing run is a bad run, if the average read quality is poor after half the read length you won't get a good assembly regardless of what you do.

If this is a PE read run you might want to check quality of forward and reverse reads and likely drop the reverse reads completely if they are considerably bad.

I used trimmomatic to trim my Miseq PE reads. If adapter sequences have not been trimmed, trimmomatic can do this while quality trimming and filtering provided you know what adapters were used.

Use the --careful flag when running spades.

The fragmented RNAs error is coming from all the small 'junk' contigs that are short usually low coverage. They arise because of multiple rRNA gene copies in the genome that can't be resolved by the assembler without a scaffold from mate paired or super long reads. I filter out these short low coverage reads after assembly using a separate script.

The adapter contamination can happen even after pre quality trimming and filtering and be corrected by locating the contig in question and manually trimming off the adapter sequence.

Hi Qinhong

Some points:

- At the first step, it is necessary to recheck the read quality and average coverage, to ensure the use of high-quality reads.

- Also, it is difficult to choose the best assembler for bacterial genomes, so you can try some others instead of SPAdes such as REAPR or FRCBam, But also I prefer the SPAdes. This link can help you: www.cbcb.umd.edu/software/imetamos.

- To remove the adaptors, I suggest to re-trimming.

- Moreover, you can have a look at enclosed papers.

I hope it will be useful

Thank you all so much for the kind help!!

The attach please find a fastQC plot of the original file. Is it too bad to start with?

They used 300 cycle Miseq kit V2 to generate library and we have 2,260,831 reads. The final assembly was  3,139,682 bp. I am not sure if it means the coverage = 300*2,260,831/3,139,682=216X... When they run the illumina they set the required coverage as 100X

I am really new to this field. Hope it dose not sound too silly.

Thank you Artur and everyone so much for the kind help! I will try the more aggressive trimming and experiment with the seembler/assembly parameters :)

Trimming with trimmomatic is pretty good. Apart from trimming though, have you considered error correction strategies. Error correction is important and helps to improve genome assembly. From the file you provided, you seem to have really good qualities but there maybe some errors (substitution) with illumina platforms. Some assemblers also perform error correction and you may look at those too.

Dear Issac,

Thank you for you kind response. Would you please recommend some of these assemblers? Thanks a again.