Paired end seq read lengths are different

[statsteam]

Hi all,

I recently acquired a dataset from GEO (HiSeq 2500, accession: GSE107029). It is a paired-end but the read_1 is 94 bp while read_2 is 100 bp. Since I've never seen paired-end data with different read length for read_1 and read_2 from HiSeq 2500, I am wondering if anyone can help me understand why read_1 and read_2 have different read lengths.

Here are a few reads from read_1 and read_2. I downloaded data using
fastq-dump --split-files SRR6300667

Read_1 (SRR6300667_1.fastq)
@SRR6300667.1 DHCDZDN1:3:1101:1145:1177 length=94
CGGAATGCAGCAATCAATGTCGTCGGAAGATCCTGAATAAATCCTACTGTATCTGAAAGAAGAACACTGTAGCCGCTTGGCAGGACCATTTTTC
+SRR6300667.1 DHCDZDN1:3:1101:1145:1177 length=94
DFDHHH<EEFGGIHIIHCGFDGDF@GHFADFHGICFAEHGHECCAG@GG;EH>CCEA73?;B>@CCCCCCCCCCBBBB???C?BB@@?CCEEC#
@SRR6300667.2 DHCDZDN1:3:1101:1178:1247 length=94
GGCTCCCCCCTGCAAATGAGCCCCAGCCTTCTCCATGGTGGTGAAGACGCCAGTGGACTCCACGACGTACTCAGCGCCAGCATCGCCCCACTTG
+SRR6300667.2 DHCDZDN1:3:1101:1178:1247 length=94
FHHHHHJJJJJJJIIJJJJJIJJJJJIJJJJJJIJJJJGHHAEHIHIIHHFFCEEEEDDDDDDDDDDDABDDDDDDDDBDDDDDBDDDDDDDD@
@SRR6300667.3 DHCDZDN1:3:1101:1313:1046 length=94
TCCTTTAGCTGACCACTTCTTCAAGTAGGCCGGGGATACAAAATCCTTTTGCATGAGGAAAGCTGAAATTCCACACAGGTACCACAAGATATTA
+SRR6300667.3 DHCDZDN1:3:1101:1313:1046 length=94
EHHHHHEGBGGCHIJGHIFHIHIIIIIIJJJHIIJAHGFGIIJJCFGGGIIBCHHEHGFDEFFEEECCCEDCCCCBDDD:@CCACBBDCDDEED


Read_2 (SRR6300667_2.fastq)
@SRR6300667.1 DHCDZDN1:3:1101:1145:1177 length=100
CGATGACCAGAAAAATGGTCCTGCCAAGCGGCTACAGTGTTCTTCTTTCAGATACAGTAGGATTTATTCAGGATCTTCCGACGACATTGATTGCTGCATT
+SRR6300667.1 DHCDZDN1:3:1101:1145:1177 length=100
@<ADDDDHBHFFEGGGE<CFGHIIIIGCEGDHIGI@GGGCFGHIIIIIHCHAGGHIG@@D>DGHGCACAEEHDFFFFFEDA>B@;,5@3>ADC:A@CCC:
@SRR6300667.2 DHCDZDN1:3:1101:1178:1247 length=100
ATGTTCCAATATGATTCCACCCATGGCAAATTCCATGGCACCGTCAAGGCTGAGAACGGGAAGCTTGTCATCAATGGAAATCCCATCACCATCTTCCAGG
+SRR6300667.2 DHCDZDN1:3:1101:1178:1247 length=100
CCFFFFDHHHDADEHGGGJJJEECHGDFHGIIJCDGHIGIJJFGAHEHGGGHGBHGEHIIIGHFHEHDDDD@EACECEECDDCC>CACD<>CDCCDCCD9
@SRR6300667.3 DHCDZDN1:3:1101:1313:1046 length=100
AGCCATACAGGAGATGGGAAACCACGCTATGATACTTTCTGGAAACATTTTATATTTGTTATGATGGACATTTTGCTCGATTGGAGCATGCATAATATCT
+SRR6300667.3 DHCDZDN1:3:1101:1313:1046 length=100
BCFFFFFHHHHHIHIIIJGJFGHIJJIJJJJIIGGHIJIJJJFJIAHHIHHIJIIJJJJJGIJJIJJJIGIHHHHEHFFFEECECDA?CCDDDDCDDEEF

I am quite confused.

Thank you,
Statsteam

[GenoMax]

It is possible that the sequencer may have had trouble with last 6 cycles on read 1 so that part of the data has been trimmed. That can be one explanation. You should be able to use this data without any issues.

[nucacidhunter]

There is no technical constraint on the length of Illumina paired-end reads apart from R1 which should be 25 cycles. In some applications using asymmetric read length is more cost effective. For instance, 10x Genomics single cell RNA-seq libraries can be sequenced in 2x100 configuration using 200 cycle sequencing kit but it can be sequenced 28 cycles for R1 and 90 cycles for R2 using 100 cycle kit with identical outcome.