Erroneous repeat resolution

[mjsull]

Description of bug

Hi,

We are having some problems with our assemblies. Spades is resolving a single long contig in the before_rr.fasta into itself and its reverse compliment back-to-back. This shouldn't be possible with paired-end illumina reads. BLAST tab output of the contig below.

NODE_1_length_166755_cov_35.885210      NODE_1_length_166755_cov_35.885210      100.000 166755  0       0       1       166755    1       166755  0.0     3.079e+05
NODE_1_length_166755_cov_35.885210      NODE_1_length_166755_cov_35.885210      99.999  83327   0       1       83429    166755  83326   1       0.0     1.539e+05
NODE_1_length_166755_cov_35.885210      NODE_1_length_166755_cov_35.885210      99.999  83327   0       1       1       83326    166755  83429   0.0     1.539e+0

This isn't an issue in Spades 3.X

spades.log

program completed successfully.

params.txt

Typically we run with --isolate mode, but happens with default setting too. FLASH to merge overlapping paired-end reads improved the issue but didn't fix it.

SPAdes version

4.X

Operating System

ubuntu

Python Version

3.8.10

Method of SPAdes installation

conda

No errors reported in spades.log

• Yes

[asl]

Nothing changed in repeat resolution in 4.x vs latest in 3.x series. In any case, in order to reproduce the issue we will need original reads as well as exact cmdline.

[mjsull]

Hi @asl, this is a clinical sample so can't share the reads publicly. Happy to share privately. Do you mind sending an email to mjsull@gmail.com and I will send you a drive link.

The command run was

spades.py --isolate -1 reads_R1.fastq.gz -2 reads_R2.fastq.gz -o spades_4.0_output

[mjsull]

Hi @asl, digging into this a little further, spades resolves 1571769+,1336428+, 1568579-,1571769- into a single ~160kbp contig. While this makes sense from a graph traversal perspective, I think this graph is the result of 2 conformations of a molecular switch in the bacterial genome and don't represent a state the genome actually exists in. The result is large sections of duplicated genome.