25% of reads too short Unmapped in SmartSeq2

[Rohit-Satyam]

Dear @alexdobin

Greetings!!

I am running Starsolo on SmartSeq-2 data for my parasite data obtained from GSE145080 and the reads are 71 bp long PE. I carried out the ribodepletion using ribodetector and then used the non-rRNA reads for alignment. I didn't remove adapters I use the following parameters:

## indexing
STAR-avx2 --runThreadN 10 --runMode genomeGenerate --genomeDir 00_index \
--genomeFastaFiles ToxoDB-68_TgondiiME49_Genome_withERCC.fasta --sjdbGTFfile ToxoDB-68_TgondiiME49_withERCC.gtf \
--genomeSAindexNbases 12 --genomeChrBinNbits 15 --sjdbOverhang 70

STAR --genomeDir !{index.toRealPath()} --runThreadN !{task.cpus} \
--alignIntronMax 1000000 \
--alignIntronMin 20 \
--alignMatesGapMax 1000000 \
--alignSJDBoverhangMin 1 \
--alignSJoverhangMin 8 \
--genomeLoad NoSharedMemory \
--outBAMsortingThreadN !{task.cpus} \
--outFileNamePrefix !{sid}_ \
--outFilterMatchNminOverLread 0.3 \
--outFilterMismatchNmax 999 \
--outFilterMismatchNoverReadLmax 0.04 \
--outFilterMultimapNmax 20 \
--outFilterScoreMinOverLread 0.3 \
--outFilterType BySJout \
--outSAMattributes All \
--outSAMtype BAM SortedByCoordinate \
--outSAMunmapped Within \
--outReadsUnmapped Fastx \
--outSAMattrRGline ID:${id} SM:!{sid} PL:ILLUMINA \
--readFilesCommand zcat \
--readFilesIn !{reads[0]} !{reads[1]} \
!{params.staralign_ext}\
--outBAMsortingBinsN 200 \
--sjdbGTFfile !{params.gtf}  \
--sjdbScore 1 \
--soloType SmartSeq \
--outSAMstrandField intronMotif \
--soloUMIdedup Exact NoDedup \
--soloStrand Unstranded --soloFeatures Gene GeneFull SJ \
--soloCellFilter None \
--soloOutFileNames output/ features.tsv barcodes.tsv matrix.mtx \
2> !{sid}.stderr

The too-short read percentage is same across almost all the cells. When I check these reads, these are mostly 28S rRNA from Toxoplasma (same organism) reads. But that's no excuse not to map.

                                 Started job on |	Dec 28 16:14:34
                             Started mapping on |	Dec 28 16:14:40
                                    Finished on |	Dec 28 16:16:54
       Mapping speed, Million of reads per hour |	16.81

                          Number of input reads |	625651
                      Average input read length |	142
                                    UNIQUE READS:
                   Uniquely mapped reads number |	336871
                        Uniquely mapped reads % |	53.84%
                          Average mapped length |	124.96
                       Number of splices: Total |	35148
            Number of splices: Annotated (sjdb) |	34259
                       Number of splices: GT/AG |	34867
                       Number of splices: GC/AG |	268
                       Number of splices: AT/AC |	5
               Number of splices: Non-canonical |	8
                      Mismatch rate per base, % |	1.08%
                         Deletion rate per base |	0.02%
                        Deletion average length |	1.46
                        Insertion rate per base |	0.01%
                       Insertion average length |	1.45
                             MULTI-MAPPING READS:
        Number of reads mapped to multiple loci |	62649
             % of reads mapped to multiple loci |	10.01%
        Number of reads mapped to too many loci |	2870
             % of reads mapped to too many loci |	0.46%
                                  UNMAPPED READS:
  Number of reads unmapped: too many mismatches |	0
       % of reads unmapped: too many mismatches |	0.00%
            Number of reads unmapped: too short |	181133
                 % of reads unmapped: too short |	28.95%
                Number of reads unmapped: other |	42128
                     % of reads unmapped: other |	6.73%
                                  CHIMERIC READS:
                       Number of chimeric reads |	0
                            % of chimeric reads |	0.00%

[Rohit-Satyam]

Here is the multiqc report for RH_96 well plate. The also mentioned in their paper about this but didn't offer any explanation:

We reported ‘% mapped’ based on the meta-alignment output from STAR aligner. We checked for some of the unmapped reads on BLASTn and found the majority of them to map to Toxoplasma 28S ribosomal RNA. 

multiqc_report.html.gz