Sailfish quant returning Boost exceptions

47 views
Skip to first unread message

Robert Zinna

unread,
Sep 25, 2015, 4:01:40 PM9/25/15
to Sailfish Users Group
Hello! I'm a graduate student trying to get a handle on different quantification methods for RNAseq data. I've successfully built sailfish indices for my transcriptome, but I'm running into problems when I try and use sailfish quant.

The command I'm running is the following

sailfish quant -i <path to index> -l "T=PE:O=><:S=U" -1 left -2 right -o <path to output>

in the above command "left" and "right" refer to files containing a list of files of reads, as seen below. 

less left
tissue_1_R1.fastq
tissue_2_R1.fastq
tissue_3_R1.fastq

however, when I run sailfish quant, I get the following error message. 
================
Exception : [boost::bad_any_cast: failed conversion using boost:any_cast]
================
sailfish quant was invoked improperly.
For usage information, try sailfish quant --help

For reference, I'm using the older version (0.6.3), because I'm working on a shared computer cluster, and they haven't updated.

My question is this- is the error message I'm receiving above due to me improperly inputting sailfish commands, or is this something I need to take up with my Sysadmins on the cluster, to update Sailfish and look into the Boost errors?

Thanks in advance,
Robert

Rob

unread,
Sep 25, 2015, 4:37:20 PM9/25/15
to Sailfish Users Group
Hi Robert,

  Welcome to the user group!  So, let me preface this by saying that the newer version of Sailfish contains significant improvements to both speed and accuracy (i.e. it's not just a bugfix release, but it substantially improves some of the underlying methodology).  So, I highly recommend you have the sysadmins on the cluster upgrade to the newest version.  You'll be able to build indices more quickly, the resulting indices will be smaller, you'll be able to quantify considerably faster, and the results will be, in general, more accurate.

  With that out of the way, I believe the error you're seeing is a result of the way you're providing the input.  In particular, you shouldn't provide a file containing a list of the files you want to quantify, but rather, you provide the file names to quantify directly on the command line.  For example, in your case it would be something like the following:

  sailfish quant -i <path to index> -l "T=PE:O=><:S=U" -1 tissue_1_R1.fastq tissue_2_R1.fastq tissue_3_R1.fastq -2 tissue_1_R2.fastq tissue_2_R2.fastq tissue_3_R2.fastq -o <path to output>

As an aside, are you certain that you want to quantify these samples together?  If they actually are from e.g. different tissue types, and your plan is to subsequently try to look at differences in expression between these tissues, then you probably want to quantify them separately.  Providing multiple fastq files on the command line will treat them as coming from the same sample, and you will get one quantification file out that is the result of quantifying with all of these reads together.

Best,
Rob

Robert Zinna

unread,
Oct 27, 2015, 4:34:59 PM10/27/15
to Sailfish Users Group
Rob-

Thanks for your quick reply! Sorry I took so long to respond in kind. I'm working with our sysadmins to get the newest Sailfish installed. 

I didn't realize that providing all of the files in a list would treat them as the same sample- I definitely want to quantify them separately. I assumed that the output would be sorted by input file, and I'm glad you warned me before I made a mistake! 

I will note though, that I did try to run an analysis on only two tiles (left and right read inputs), and received the same error message. 

For example, I ran the command 

sailfish quant -i <path to index> -l "T=PE:O=><:S=U" -1 tissue_1_R1.fastq -2 tissue_1_R2.fastq -o <path to output>

and got the same error as above. Does this make any sense? Is it undiagnosable until I update the software?

Thanks in advance,
Robert
Reply all
Reply to author
Forward
0 new messages