Explain "Error occurs when reading inputs"

[jan-glx]

Megahit fails for me with "Error occurs when reading inputs". It seems to work with a subset of the reads. Is this some kind of memory overflow?

2020-04-24 01:19:23 - MEGAHIT v1.2.9
2020-04-24 01:19:23 - Using megahit_core with POPCNT and BMI2 support
2020-04-24 01:19:23 - Convert reads to binary library
2020-04-24 01:19:23 - command /.../MEGAHIT-1.2.9-Linux-x86_64-static/bin/megahit_core buildlib /.../2_S1/tmp/reads.lib /.../tmp/reads.lib   
2020-04-24 01:19:53 - b'INFO  sequence/io/sequence_lib.cpp  :   77 - Lib 0 (/.../2_S1_R1.fq.gz,/.../2_S1_R2.fq.gz): pe, 18314632 reads, 151 max length'
2020-04-24 01:19:55 - b'INFO  utils/utils.h                 :  152 - Real: 31.4624\tuser: 19.6492\tsys: 2.9616\tmaxrss: 263564'
2020-04-24 01:19:55 - Error occurs when reading inputs

[s4251484]

Hi there i encounter the same error too!

But I cant figure out how to resolve it. @ctb @lgautier @sebhtml @aquaskyline please enlighten!

[s4251484]

Hi there i encounter the same error too!

But I cant figure out how to resolve it. @ctb @lgautier @sebhtml @aquaskyline please enlighten!

my problem is resolved by regenerating one fastq file - likely was corrupted during the one of the file-transferring process. hope this helps someone! cheers!

[erikrikarddaniel]

I'm getting the same error, when the number of libraries gets over a certain number (v. 1.2.9; I tried both a Conda package running in a Singularity container and a computing-center provided module). For me, it doesn't help to concatenate the fastq files as it did for @s4251484. Nor does it help to allocate more memory. Going from a machine with 128 GiB to one with 256 GiB results in the same failure for the same libraries. I'm happy to share data for debugging.

[ntromas]

Hi,

Wonder if anyone found a solution for this?

Nico

[USMortality]

I encountered a similar error with NC_038311, which contains a W in the sequence, which seems to trip megahit.

cargo install --git https://github.com/stjude/fqlib.git   

I can recommend to lint your input files, e.g.: fq lint out/SRR00000001_1.fastq out/SRR00000001_2.fastq

[ilasadar]

This experience might provide a solution for others. I initially faced the same problem when I used Trimmomatic to trim the reads. I concatenated the unpaired reads (those where forward or reverse reads did not survive) with the paired reads (concatenating was a mistake), then attempted to run MEGAHIT, resulting in the same error. However, when I processed only the paired reads, MEGAHIT ran without any issues. So, consider what you had done before the megahit run.

[vinisalazar]

Hi, just coming to report I am facing the same problem, I performed QC and trimming with Fastp, host removal with Minimap2, paired reads with fastq-pair to ensure same amount of R1 and R2 reads, and then I get this error for about half of my samples. The other half works fine. I'll try @USMortality's linting solution to see if the problem is ambiguous bases. I'm still not sure if the problem is in my upstream steps (sure doesn't look like it) or if it's just a megahit problem.

[jolespin]

Anyone ever figure this out?

[Li-Alvarez]

Hi! I had the same issue, I solved it by decompressing the fastq.gz file to fastq. I hope this helps.