Trinity: option to increase bowtie memory allocation

[ataraxia]

Hi all,
I have pair-end data that I am currently analysing using Trinity.

I am now running the following script provided by Trinity to estimate abundance:

run_RSEM_align_n_estimate.pl --seqType fq --SS_lib_type RF --left seq1.fq --right seq2.fq --thread_count 16 --prefix /output/RSEM --transcripts Trinity.fasta

and although I get all the output files, I get errors in the log:

Warning: Exhausted best-first chunk memory for read X; skipping read
(or around 11000 reads)

# reads processed: 188565323
# reads with at least one reported alignment: 82758840 (43.89%)
# reads that failed to align: 105806483 (56.11%)
Reported 225985477 paired-end alignments to 1 output stream(s)
[bam_sort_core] merging from 335 files...


From my understanding, you can play with the following setting in Bowtie (--chunkmbs) to increase memory requirements for alignment.

But is there any possibility to do so using the perl script provided in Trinity (run_RSEM_align_n_estimate.pl)? Or should I analyse my data with RSEM directly, as it offers a --bowtie-chunkmbs option?

[lzsph]

UPDATED:

Go to the trinityrnaseq_r2013-02-05/trinity-plugins/rsem folder, find a script named "rsem-calculate-expression", open it with a text editor, at line 30, there is

my $chunkMbs = 0;# 0 = use bowtie default

, change the "0" to "200" or other integer, then save it. It means it add the --chunkmbs in the command when you run the

run_RSEM_align_n_estimate.pl

, I ran the script again after edit the script, it appeared like this:

bowtie -q --phred33-quals -n 2 -e 99999999 -l 25 -I 1 -X 1000 --chunkmbs 200 -p 4 -a -m 200 -S TRANS -1 Ms-1-1_R1.fastq -2 Ms-1-1_R2.fastq | /Users/lzsph/scripts/trinityrnaseq_r2013-02-25/util/RSEM_util/../../trinity-plugins/rsem/sam/samtools view -S -b -o RSEM.temp/RSEM.bam -
[samopen] SAM header is present: 192875 sequences.
# reads processed: 20000000
# reads with at least one reported alignment: 16275899 (81.38%)

No warning message appeared.


Before I edit the script mentioned above, I also found this QA here (http://sourceforge.net/mailarchive/f...tyrnaseq-users), they said re-run the program may solve this problem. But when I reran the perl script, still the same problem, maybe my personal computer has a lower RAM (Mid-2011 iMac, 8GB RAM). I also Googled a lot, so I want to add the --chunkmbs in the script.

Hope it helps.

Regards,
-s

[ataraxia]

Thanks lzsph,
This fixed my problem.