Low Read Maps to Scaffolds using MEGAHIT

User Support
1

Hello, I am new to bioinformatics analysis, and I am currently running a set of complex soil samples via qiime2-moshpit-2025.7 installed via conda. I am posting to hopefully resolve an issue with my assembly generating a low percentage of map reads to scaffold (34%) and a low-quality binning (2.44%). I will link a table that outlines my diagnostic percentages. I am currently using the default megahit and metaBAT parameters (as seen on the moshpit tutorial pdf) for assembly and binning. I am wondering whether co-assembly and binning of replicates would increase this, or whether I should adjust my assembly and binning parameters. Ex. using meta-large or meta-sensitive

DiagTable

My workflow is as current workflow is as follows:

  1. demux
  2. cutadapt-trimmed paired
  3. megahit assembly
  4. index contigs
  5. map reads to contigs
  6. bin contigs with metaBAT
  7. bin evaluation with BUSCO

I have more downstream analysis planned, but testing to get an optimized assembly prior.

Any advice or assistance would be greatly appreciated. My apologies if I have misused the forum in any way. I cannot currently identify any troubleshooting tips for this situation.

5

Hello @MasonDavis , welcome to the Forum!

@SoilRotifer Suggests to update to the latest moshpit version and then use fastp instead of cutadapt. He says he observed that far more reads map to genomes (for both host removal, and microbial mag mapping) after fastp processing.

You can have a look at the new moshpit fastp docs for this.

Let us know if this helps! :slight_smile:

7

Hello @Vinzent_Risch, thank you for the reply. I was able to diagnose my issue via a sylph and nonpareil analysis, which revealed that my samples were highly diverse, with a median sample coverage of ~0.3 and an average taxa percentage just under 3%. Utilizing fastp did not assist with my assembly, but what did end up giving me better coverage (increased mean contig lengths, map reads and properly paired by ~4%) was removing the 21 and 141st k mer from megahit for a single sample assembly. I am moving now to the co assembly of 4 replicates, to try and recover as much as possible. I appreciate the advice.

Thank you!

Mason Davis