ema-bin-??? and possible preproc failure about ema HOT 14 OPEN

Hi,
Yes, ema preproc should produce 500 (-n) files of the form ema-bin-000, ema-bin-001 etc. (the ??? is a bash idiom for matching files of this form). I wonder if there is some quirk with that FASTQ file -- could you post the first 10 or so lines of it?

from ema.

from ema.

That FASTQ looks good -- how do the *.ema-ncnt files look? Are they produced by ema count?

from ema.

No clue how the .ema-ncnt files really look. They are not text. I have only two files: counts_file.ema-ncnt and counts_file.ema-fcnt. The following is the first 10 lines of counts_file.ema-ncnt:
� *
9���
-���
��
����
�
����
���
���
��c�

The log for the count is:
:: Loading 10X took 2.4 s
:: Dumped block 1
:: Dumped block 2
:: Dumped block 3
:: Dumped block 4
:: Dumped block 5
:: Dumped block 6
:: Dumped block 7
:: Dumped block 8
:: Counting took 2472.2 s
:: Reads with OK barcode: �Ò637,148,268 out of Ð683,891,843
:: Ignored th OK barcode: �Ò637,1480 reads
:: Dumped block 9
:: Printing took 8.6 s
:: Processed oo683,891,843 reads ( 470,201 MB uncompressed) in 2,483 s

--Thanks for your help!

from ema.

Hi @michaeljmetzger

Can you please try doing this instead:

cat ~/EMA/some_clam_S3_L004_interleaved.fastq | ema preproc -w ~/EMA/4M-with-alts-february-2016.txt -n 500 -t 40 -o test *.ema-ncnt 2>&1 | tee preproc.log

Istest directory empty or not? I have a feeling that passing -o ~/EMA confuses the script because of ~.

from ema.

from ema.

So the log contains all these non-ASCII characters? Weird... any chance you can share the first 10k records from the interleaved file so that I can take a look?

from ema.

from ema.

OK, that is certainly unexpected.

Which OS are you using? You can email me the data link (inumanag at mit dot edu--- first 10k reads should be fine) and I can take a look. Currently, we don't have any small sample for these purposes.

from ema.

from ema.

Hi @michaeljmetzger

Just following up on this one: did you manage to resolve this issue with another compiler or pre-compiled binaries?

from ema.

from ema.

Hi @michaeljmetzger

Yeah, I believe that there is something wrong with the compiler on your cluster because the generated binary is completely corrupted--- probably because some compiler library versions got mixed up (this happens sometimes on cluster environments).

As for Long Ranger vs. EMA: yeah, they are. We found out that EMA's results are more robust, esp. in repetitive regions, so we always map with EMA.

from ema.

I have now given it one more shot and installed using conda. I can see the help text when I type "ema help" but the command yields a completely empty log file and no count files at all and no errors shown.

cat ~/EMA/some_clam_S3_L004_interleaved.fastq | ema count -w ~/EMA/4M-with-alts-february-2016.txt -o counts_file 2>&1 | tee count.log

Any thoughts?

from ema.