Num. reads assembled: 1669184
Avg. total coverage: 16.60
Number of contigs: 13991
Largest contig: 112360
N50 contig size: 1171
that is the summary I got out of a de novo assembly (ion torrent data and MIRA3 assembler
I would like a suggession as to how you would rate the sequence.
Bacterial genome, expected to have a few plasmids.
How will you rate the quality of this sequence?
I have tried it with Newbler, which are like
numAlignedReads = 1647777,
N50ContigSize = 955,
numberOfContigs = 26930,
largestContigSize = 129390;
what do I do with these now... suggestions required.
MIRA has been said to be a good assembly for Bacteria (with small genome). However, its assembly improves so much doing an extra assembly with CAP3 software. MIRA + CAP3 combination is the best assembly described for bacterial genomes.
In the case you describe, it's obvious that MIRA has performed a better assembly than Newbler: higher N50 and less contigs, which may mean NO REDUNDANCY. For checking the redundancy I would recommend you to do a BLASTn of your contigs against themselves. In the case of MIRA, I would hope to get less hits than in the case of Newbler.
If I were you, I would choose the MIRA assembly. But a BLAST analysis will confirm you.
MIRA has been said to be a good assembly for Bacteria (with small genome). However, its assembly improves so much doing an extra assembly with CAP3 software. MIRA + CAP3 combination is the best assembly described for bacterial genomes.
In the case you describe, it's obvious that MIRA has performed a better assembly than Newbler: higher N50 and less contigs, which may mean NO REDUNDANCY. For checking the redundancy I would recommend you to do a BLASTn of your contigs against themselves. In the case of MIRA, I would hope to get less hits than in the case of Newbler.
If I were you, I would choose the MIRA assembly. But a BLAST analysis will confirm you.
Hi,
I would recommend to filter the reads using a Q30 (or Q20). Also a minimum length of 50 bp could be an useful filter. There are different tools to do that, such as Ea-utils, FastX-Toolkit or Prinseq. Increase the Q to 30 would give you a better assembly.
For the assembly, iAssembler is a pipeline that uses MIRA+CAP3 (http://bioinfo.bti.cornell.edu/tool/iAssembler/#introduction) so perhaps could improve the assembly that you have. Also you could use a De Brujin graph based assembler such as ABySS or SOAPdenovo although the read coverage is not around 40-50X. If you have pair-ends perhaps may be interesting to use SOAPdenovo, because it has a postassembly tool to fill the gaps based in a overlapping method, that usually improve the stats.
Good luck,
The quality of the sequences are rated by its quality score, the higher the quality the better. I suggest to stick to Q30 and through the shorter reads below 50bp. You have to also give a closer attention to the alignment quality, i.e the assembler setting. I suggest to start with the most strict setting and gradually change a few setting till you would be satisfied with the final result. If your sequences are more than 100bp, check the quality pattern and try to trim the bad quality regions from the beginning and end. Finally, glue them together using your favorite assembler .
That is my 2 cents.
Have fun!!!
I have experienbce with MIRA for bacterial genomes.
Mira is a really coinservative assembler. It prefer to provide more contigs than to have wrong misassembled contigs. Up to date is one of the best assembler. Something you should check is the proper assembly. For that I suggest you to install gap5 and try to see how it really looks. If you have some proble to do that or to convert your assembly to gap5 format let me know.
If you think that some plasmid or other extrachromosomal element is present you should check the contigs coverage and see if some of them go higher than average coverage. for that take a look to the contig_stats file.
Take a look at the MIRA manual and cases... Most parameters could be changed according to their description in the manual.
I hope this help.
Overall sequencing Q score was low. Q30 would be too stringent. Will loose 85% data then. AT Q20, the average length is also going down to 48%.... so I choose to Q15 as cutoff. Again, is it a problem with my sequence run or is it a general trend with Ion data. What I learned, that the difinition of Q score of ion torrent is different. It also depend on the read length. Please suggest
@Aureliano Bombarely : Could you please confirm whether iAssembler takes DNA data ? Because in the iAssembler homepage, it was written that it takes transcriptome data to do denovo asssembly !!
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools