Low alignment rate

[MarcelloMalpighi]

Hi,
I quantified a paired-end Smart-seq2 RNA-Seq dataset (SRR3936136) using kallisto v0.48.0 with the GENCODE v47 transcriptome. The fragment length was calculated by CollectInsertSizeMetrics (Picard) based on STAR bam. The alignment rate was 34.2% in paired-end mode and 45.6%/43.6% in single-end mode, compared to 74.93% obtained with STAR. I wonder why there is notable discrepancy in the alignment rate. The corresponding information is provided below.
kallisto pair-end run_info.json:

{
        "n_targets": 387944,
        "n_bootstraps": 0,
        "n_processed": 1981171,
        "n_pseudoaligned": 676920,
        "n_unique": 68993,
        "p_pseudoaligned": 34.2,
        "p_unique": 3.5,
        "kallisto_version": "0.48.0",
        "index_version": 10,
        "start_time": "Sat Dec 21 00:02:34 2024",
        "call": "kallisto quant --pseudobam --genomebam --single-overhang -l 302.431195 -s 138.561984 -i /home/Usersdata2/references/gencode/kallistoIndex/gencode.v47.primary_assembly_k15.index -o /home/Usersdata2/abundance/SRR3936136 -t 32 --gtf /home/Usersdata2/references/gencode/gencode.v47.primary_assembly.annotation.gtf --chromosomes /home/Usersdata2/references/gencode/kallistoIndex/gencode.v47.primary_assembly.annotation.chromosomeInfo.txt /home/Usersdata2/datasets/SRP079058/fastq/SRR3936136_1.fastq.gz /home/Usersdata2/datasets/SRP079058/fastq/SRR3936136_2.fastq.gz"
}

kallisto single-end run_info.json:

{
        "n_targets": 387944,
        "n_bootstraps": 0,
        "n_processed": 1981171,
        "n_pseudoaligned": 903816,
        "n_unique": 84772,
        "p_pseudoaligned": 45.6,
        "p_unique": 4.3,
        "kallisto_version": "0.48.0",
        "index_version": 10,
        "start_time": "Sat Dec 21 19:35:34 2024",
        "call": "kallisto quant --pseudobam --genomebam --single-overhang -l 302.431 -s 138.562 -i /home/Usersdata2/references/gencode/kallistoIndex/gencode.v47.primary_assembly_k15.index -o /home/Usersdata2/kallisto/testSingle/read1 -t 16 --single --gtf /home/Usersdata2/references/gencode/gencode.v47.primary_assembly.annotation.gtf --chromosomes /home/Usersdata2/references/gencode/kallistoIndex/gencode.v47.primary_assembly.annotation.chromosomeInfo.txt /home/Usersdata2/datasets/SRP079058/fastq/SRR3936136_1.fastq.gz"
}

{
        "n_targets": 387944,
        "n_bootstraps": 0,
        "n_processed": 1981171,
        "n_pseudoaligned": 862909,
        "n_unique": 81911,
        "p_pseudoaligned": 43.6,
        "p_unique": 4.1,
        "kallisto_version": "0.48.0",
        "index_version": 10,
        "start_time": "Sat Dec 21 19:38:53 2024",
        "call": "kallisto quant --pseudobam --genomebam --single-overhang -l 302.431 -s 138.562 -i /home/Usersdata2/references/gencode/kallistoIndex/gencode.v47.primary_assembly_k15.index -o /home/Usersdata2/kallisto/testSingle/read2 -t 16 --single --gtf /home/Usersdata2/references/gencode/gencode.v47.primary_assembly.annotation.gtf --chromosomes /home/Usersdata2/references/gencode/kallistoIndex/gencode.v47.primary_assembly.annotation.chromosomeInfo.txt /home/Usersdata2/datasets/SRP079058/fastq/SRR3936136_2.fastq.gz"
}

STAR Log.final.out

                                 Started job on |       Dec 21 00:31:43
                             Started mapping on |       Dec 21 00:35:30
                                    Finished on |       Dec 21 00:36:24
       Mapping speed, Million of reads per hour |       132.08

                          Number of input reads |       1981171
                      Average input read length |       128
                                    UNIQUE READS:
                   Uniquely mapped reads number |       665253
                        Uniquely mapped reads % |       33.58%
                          Average mapped length |       127.44
                       Number of splices: Total |       142473
            Number of splices: Annotated (sjdb) |       142473
                       Number of splices: GT/AG |       140613
                       Number of splices: GC/AG |       1251
                       Number of splices: AT/AC |       64
               Number of splices: Non-canonical |       545
                      Mismatch rate per base, % |       0.37%
                         Deletion rate per base |       0.01%
                        Deletion average length |       1.44
                        Insertion rate per base |       0.01%
                       Insertion average length |       1.25
                             MULTI-MAPPING READS:
        Number of reads mapped to multiple loci |       819140
             % of reads mapped to multiple loci |       41.35%
        Number of reads mapped to too many loci |       3
             % of reads mapped to too many loci |       0.00%
                                  UNMAPPED READS:
  Number of reads unmapped: too many mismatches |       552
       % of reads unmapped: too many mismatches |       0.03%
            Number of reads unmapped: too short |       494129
                 % of reads unmapped: too short |       24.94%
                Number of reads unmapped: other |       2094
                     % of reads unmapped: other |       0.11%
                                  CHIMERIC READS:
                       Number of chimeric reads |       0
                            % of chimeric reads |       0.00%

Thanks in advance.

[mschilli87]

What fraction of your STAR alignments is intergenic or overlapping introns?

[MarcelloMalpighi]

I used the following command to count reads aligned to the reference transcriptome by STAR, then subtracted it from the total read number. The result indicates that 52.96% of the reads were not derived from exons. Does this imply that the alignment rate is relatively low due to a significant proportion of reads originating from non-exonic regions?
>samtools view SRR3936136_Aligned.toTranscriptome.out.bam | awk '{print $1}' | sort | uniq | wc -l
>932000

[mschilli87]

I would assume so. Kallisto usually (pseudo)aligns to annotated spliced transcripts only. So your comparison to STAR is not really apples to apples.

[MarcelloMalpighi]

The original literature indicates that it is a standard single-cell RNA-seq library.