Hi all!
I have been trying to get a good assembly of my fungal organism (I have paired end 250bp read data), here are the steps I follow:
- adapter removal
- quality trimming of reads based on base quality
- removal of contaminating bacteria (E.coli) reads
Even after this, I somehow still see some lower coverage bacteria contamination and therefore my final assembly size is larger than I would expect for this organism. I have pretty high coverage (1400X) so I decided to normalize using bbnorm (target=100, mindepth=6), however this resulted in a lot less reads, lower N50, and a even larger assembly size (see below, M=million)
Total reads(paired) Scaffold N50 Sum
Original 21M 50,501 5M
Normalized 2.9M 17,100 7M
I am wondering if the bbnorm target value I selected is perhaps too stringent? Or perhaps there is errors in the reads that is causing this contamination to still come through and I should look into error correction? Any suggestions will be very helpful, thank you!
0 answers
No answers yet.
Log in to answer this question.
-
Bacterial genome assembly, best uses of reads at excessively high sequencing depth?
written by ian.will •
-
Why does low-coverage library give more contiguous assembly than high-coverage library?
written by Timotheus •
-
Improving existing genome with new RNA-seq data
written by Dunois •
-
Detecting Chimeric Contigs
written by KSS •
-
abyss: troubleshoot assembly
written by yp19
-
Need help to remove contaminants from genome reads
written by cecilio11 •
-
Judging best assembly
written by deepti1rao •
-
How do I analyze and annotate the results of a Bacterial De novo assembly ?
written by Paul
-
Assembly method validation - am I doing it right?
written by f2369583 •
-
Merging assemblies with GAM-NGS
written by pbigbig
Have you done binning on your original assembly? That would remove bacterial contigs after the assembly, which is still fine. Also, taxonomic partitioning can be done using BlobTools.
Though your coverage is impressive, I'd be hesitant to throw away any data before the assembly as any incorrectly mapped reads amount to lost information. I am assuming here that you don't have memory issues with this dataset since you already assembled it without filtering.
This paper may be of interest as well.
Thank you for you suggestions and the linked paper. I will attempt binning to remove bacterial contigs. It seems there are many different methods of binning (here) Do you suggest any particular method to start with for this particular problem?
I was primarily referring to binning AFTER the assembly. In addition to aforementioned BlobTools, there is nice implementation using self-organizing maps in Binning. I typically use a custom t-SNE pipeline. See the supplementary material of this paper to get an idea what kind of genomic sequence separation one can expect from the last two approaches.
Thank you, Mensur. I will give this + Blobtools a shot!