A strange size difference of fastq file

[qinhao13]

Hi, currently I'm working on a Illumina sequencing data in fastq format. I downloaded it from public available database (TCGA) and it was zipped. After unzip and trimming the size of the file is about 16G. Interesting thing comes. After I copied this file to another partition, the size of the new copy became 7.6G. The number of lines in the files, the number of reads and their length distribution are the same in the two files. So I guess the two files have the same content, the new copy is not truncated.

Moreover, when I run Tophat2/Cufflinks with 16G copy, it takes much longer time to finish and the the result looks strange. But it is quite normal with the 7.6G copy. This might not be a bioinformatics question but it's quite interesting. What happened to the file? What might be those additional size in the file?

Thanks a lot.

[dariober]

I can't tell... But one thing you can try to get some hints is:

cat -vet my_strange_reads.fq | less

This is will show you non-printable characters in the file. In a typical fastq file you shouldn't see anything new in addition to the usual alphanumeric characters and some metacharacters in the read names.

In practice, I would download again the file just to make sure something got corrupted in the process.