search

kbaseattic / assembly

An extensible framework for genome assembly.
MIT License
12 stars 14 forks source link

velvet in auto recipe does not recognize FASTA files #301

Closed levinas closed 9 years ago

levinas commented 9 years ago

This bug can be reproduced against the dev server.

$ ar-run -s 140.221.67.201 --pair forward.fa reverse.fa

On the server, SPAdes launched the assembly correctly: Command: spades.py --pe2-1 forward.fa --pe2-2 reverse.fa -t 4 ...

But velvet missed all the read files and tried to run:

Command: velveth velvet_e28ac24f-d408-4f7f-9925-f6c62280ebd3 29

However, if I launch with the -a velvet option explicitly, the job runs. Is there an issue with the wasp interpretation?

levinas commented 9 years ago

This bug affects single end libraries too.

levinas commented 9 years ago

This is probably due to bayers hammer not working for fasta files which have no quality scores. As a result, nothing gets to run. We should modify the bhammer plugin to handle fasta files gracefully.

Two options:

  1. Detect FASTA files and bypass the error correction
  2. Assign the original reads to processed_reads if error correction fails.

@cbun do we already have a FASTA detector in the plugin?

cbun commented 9 years ago

Are you sure Spades runs this correctly?

BayesHammer will run, which also uses spades.py but theoretically all the assemblers should fail.

levinas commented 9 years ago

Yes, all assemblers failed. This fixes it.