Hello, See my answers below.
On Mon, 2012-03-26 at 17:08 +0530, Sushmita Mookherjee wrote: > Hi, > > Earlier I used Ray-1.7 successfully for IonTorrent single end reads > (Total no. of reads ~4 million). Now on another data of same > sequencing technology I tried to use Ray-1.7 by using the command as > follows, where the total no. of reads ~3 million: > > mpiexec -n 25 Ray -k 21 -s ../denovo_assembly/GAT_104/ > GAT_104.fastq -o ../denovo_assembly/GAT_104/Ray_output > > But no scaffolds or contigs generated and showed "Rank 0: to deal with > the sequencing error rate, try to lower the k-mer length (-k)" as an > error. Do you have a lower amount of reads in in the unsuccessful run ? This message has been removed altogether in the upcoming Ray 2.0, meaning that no peak is required in the coverage depth distribution of k-mers in your reads. But usually, you will get a peak for bacterial genomes whereas you won't for metagenomes or transcriptomes. > Then, I changed the default k-mer length 21 to 15 and got the output. I guess you mean the "same output." > Can anybody help me to get the answer, why default k-mer length > didn't work for one data whereas in other it worked fine? > Let me explain here: Ray slices your reads in smaller sequences called k-mers. In your tests, you used 15-mers and 21-mers (sequences of length 15 and 21, respectively). Then Ray builds a de Bruijn graph from these and indexes your reads in order to find long paths called contigs. Paired reads help a lot by the way in the process of constructing long error-free contigs. In your successful dataset, the k-mer are distributed in a way that it is easy, from a global point of view, to say something like "any position of your target genome is covered, on average, by X k-mer observations." In your unsuccessful dataset, there is no peak in the global distribution, probably because you don't have that much reads. However, with Ray 2.0 release candidate 5 this is not a problem. We released 2.0-rc5 today ! http://permalink.gmane.org/gmane.science.biology.ray-genome-assembler/173 > Thanks and Regards, > > Sushmita ------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure _______________________________________________ Denovoassembler-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/denovoassembler-users
