Segemehl is also a good option, since you can specify the minimum length of the spliced fragment (-Z option) in your case and also increase the accuracy (-A) and decrease evalues for split alignments (-w).

Hi all,

Thank you for your replies. Finally I've tried STAR and Bowtie (not Bowtie2).

1- I've used STAR by setting the more important parameters as follows:

--outFilterScoreMinOverLread 0.4 --outFilterMatchNminOverLread 0.4  --outFilterMatchNmin 15

Here is the final log file content generated by STAR:

                       Number of input reads |       5845995
                  Average input read length |       27
                                UNIQUE READS:
               Uniquely mapped reads number |       446299
                    Uniquely mapped reads % |       7.63%
                      Average mapped length |       25.10
                   Number of splices: Total |       4043
        Number of splices: Annotated (sjdb) |       622
                   Number of splices: GT/AG |       3930
                   Number of splices: GC/AG |       108
                   Number of splices: AT/AC |       5
           Number of splices: Non-canonical |       0
                  Mismatch rate per base, % |       1.74%
                     Deletion rate per base |       0.08%
                    Deletion average length |       1.36
                    Insertion rate per base |       0.04%
                   Insertion average length |       1.07
                         MULTI-MAPPING READS:
    Number of reads mapped to multiple loci |       4346149
         % of reads mapped to multiple loci |       74.34%
    Number of reads mapped to too many loci |       963965
         % of reads mapped to too many loci |       16.49%
                              UNMAPPED READS:
   % of reads unmapped: too many mismatches |       0.00%
             % of reads unmapped: too short |       0.31%
                 % of reads unmapped: other |       1.22%
                              CHIMERIC READS:
                   Number of chimeric reads |       1503
                        % of chimeric reads |       0.03%

According to this, I got a very low rate of uniquely mapped reads (I used the parameter --outFilterMultimapNmax 10).

How can I improve the uniquely mapped reads rate ?

2 - I've also used Bowtie to map the same sample to the same reference genome. Bowtie didn't output the number of uniquely mapped reads and the number of multimapped reads as STAR did. I checked the log file generated by Bowtie:

    # reads processed: 5845995
    # reads with at least one reported alignment: 4206928 (71.96%)
    # reads that failed to align: 1639067 (28.04%)
    Reported 98304317 alignments to 1 output stream(s)

I want to output the number of uniquely mapped reads and multimapped reads from the sam file generated by bowtie. So I have done the following:

samtools view -Sb  myfile.sam > myfile.bam 
samtools view -F 4 myfile.bam | grep -v "XS:" | wc -l

But it didn't work for me, and I got a wrong number of uniquely mapped reads. I know that those commands work for files generated by bowtie2.

Do you have any solution able to retrieve the number of uniquely mapped, and multimapped reads from the sam file generated by bowtie ?

Thanks a lot