search

iqbal-lab / Mykrobe-predictor

Antibiotic resistance predictions in minutes on a laptop
Other
50 stars 19 forks source link

Mykrobe hangs forever on BAM of 80k #23

Closed Phelimb closed 9 years ago

Phelimb commented 9 years ago

./Mykrobe.predictor.staph --file /Net/wombat/dipro/mmm/ana/Saur/PHE_Saur_for_typing/H113120068-138-1_PHE_Saur_for_typing_PHE_Saur_for_typing_v3_stampy_R00000022.bam --install_dir ~/myKrobe-predictor --format JSON

Doesn't finish or throw an error despite the fact that the bam is only 80k

-bash-4.1$ du -h /Net/wombat/dipro/mmm/ana/Saur/PHE_Saur_for_typing/H113120068-138-1_PHE_Saur_for_typing_PHE_Saur_for_typing_v3_stampy_R00000022.bam 80K /Net/wombat/dipro/mmm/ana/Saur/PHE_Saur_for_typing/H113120068-138-1_PHE_Saur_for_typing_PHE_Saur_for_typing_v3_stampy_R00000022.bam

iqbal-lab commented 9 years ago

I wonder if it is a valid bam

Phelimb commented 9 years ago

Very possibly not. In either case we should probably throw an error.

Phelimb commented 9 years ago

We should maybe have a minimum amount of data too. We are making calls on the other bams that are ~100kb despite the fact there's definitely not enough data to make a sensible call.

iqbal-lab commented 9 years ago

Well, I think this should come automatically from our confidences (plus error rate)

iqbal-lab commented 9 years ago

But yes, for now a minimum coverage of 10x say, as we only support Illumina data for now, is sensible.

iqbal-lab commented 9 years ago

Yes, actually I want to go through the code to improve the throw/catch errors. We want a clean error put in the JSON, not to die().

Phelimb commented 9 years ago

@iqbal-lab Should we fix this pre-submission?

iqbal-lab commented 9 years ago

Is fixed/gone

On 12 January 2015 at 11:25, Phelim Bradley notifications@github.com wrote:

@iqbal-lab https://github.com/iqbal-lab Should we fix this pre-submission?

— Reply to this email directly or view it on GitHub https://github.com/iqbal-lab/myKrobe-predictor/issues/23#issuecomment-69557047 .

iqbal-lab commented 9 years ago

Oh, you mean fix the making a call on too little data? I would expect to get Ns or low confidence with too little data (species code not ideal for this right now) Not for submission

On 12 January 2015 at 13:30, Zamin Iqbal zam.iqbal.lab@gmail.com wrote:

Is fixed/gone

On 12 January 2015 at 11:25, Phelim Bradley notifications@github.com wrote:

@iqbal-lab https://github.com/iqbal-lab Should we fix this pre-submission?

— Reply to this email directly or view it on GitHub https://github.com/iqbal-lab/myKrobe-predictor/issues/23#issuecomment-69557047 .

iqbal-lab commented 9 years ago

This bug is fixed