Help!, strange fastq format

illuminaGA · 05-19-2017, 10:04 AM

Dear All

I download some fastq reads from SRA, but the format is wired. The reads are all number and dots.
Please give me some pointers if you have seen this before? Thank you in advance.

example:

Code:

@SRR1175538.1 1_11_419_F5-RNA length=35
G.12.1.21.2.20.1.3013..321..20233033
+SRR1175538.1 1_11_419_F5-RNA length=35
!!@@!@!@@!@!@@!@!@@@@!!@@@!!@@@?4<@/
@SRR1175538.2 1_12_62_F5-RNA length=35
G.02.0.01.2301.1.1222.1333.111102223
+SRR1175538.2 1_12_62_F5-RNA length=35
!!@@!@!@@!@@@=!@!@@@@!@@2@!@?>?@@@@@
@SRR1175538.3 1_12_580_F5-RNA length=35
G.10.0.13.0032.2.3123.1121.022330003
+SRR1175538.3 1_12_580_F5-RNA length=35
!!<@!@!@?!@@@8!@!@2@@!@;@@!@;@@@2@@@
@SRR1175538.4 1_12_1917_F5-RNA length=35
G.00.2.12.3.10.1.1133.2333.100133010
+SRR1175538.4 1_12_1917_F5-RNA length=35
!!@@!@!@@!@!@@!@!@@@@!@@?@!@@@@@@@@@
@SRR1175538.5 1_13_1110_F5-RNA length=35
G132.2120.22320023122.3200.320102100
+SRR1175538.5 1_13_1110_F5-RNA length=35
!@@@!@?@A!5=833052/3-!@?.2!@-55/@6.?
@SRR1175538.6 1_13_1767_F5-RNA length=35
G002.3231.03320230131.1020.131301132
+SRR1175538.6 1_13_1767_F5-RNA length=35
!@@@!@@8@!@@//76*?<0@!@20@!@62/@@=-?
@SRR1175538.7 1_14_646_F5-RNA length=35
G232.1032.21300133222.2022.233022033
+SRR1175538.7 1_14_646_F5-RNA length=35
!@@@!@@@@!@@@@<@@@@@@!@@@/!@@;@@@/=@
@SRR1175538.8 1_15_1325_F5-RNA length=35
G331.2201.30111132130.0011.202011230
+SRR1175538.8 1_15_1325_F5-RNA length=35
!@@=!@*@@!@@@@@@@@@@>!@@@@!@@@?@@@?@
@SRR1175538.9 1_15_1850_F5-RNA length=35
G110.3000.03303201101.1013.001113220
+SRR1175538.9 1_15_1850_F5-RNA length=35
!:@0!9:@*!=2;.6/6@.68!@*0:!@*.@;@.3:

mastal · 05-19-2017, 10:20 AM

It's data from the SOLiD platform.

They use colorspace rather than basespace,
so you get a sequence of 0,1,2,3 rather than A,T,G,C.

The . indicate missing base calls.

illuminaGA · 05-19-2017, 11:01 AM

Thank you.

Quote:

Originally Posted by mastal

It's data from the SOLiD platform.

They use colorspace rather than basespace,
so you get a sequence of 0,1,2,3 rather than A,T,G,C.

The . indicate missing base calls.

pmiguel · 05-19-2017, 11:13 AM

If you want to map those to a reference:
(1) Don't. Delete the file and pretend like it never existed.
(2) No, really, I mean it. Delete the file.
(3) Okay, the reason there are so many "." in the sequence is that the SOLiD gave you reads from every bead it found, no matter how bad the data was. Also the .fastq's started at the very top of the slide, near the edge, where the data is worst.
(4) But you don't need to know this, right? Please tell me you took my advice and deleted the file?
(5) I mean, you can't just take the numbers and convert them to sequence. Any string of numbers could encode any of 4 sequences. It is the base at the beginning of each sequence that tells you the right conversion path. But if any "base" in the read was incorrect, it corrupts the conversion path. So if you did map these, you would want to use an old-time mapper in "colorspace" mode to do the mapping. BWA and Bowtie probably can still do this type of mapping.
(6) But you don't need to know that because you deleted the file, right? If not, well, I did warn you...

--
Phillip

Brian Bushnell · 05-19-2017, 11:31 AM

Haha

But, yes, pmiguel is completely correct and I suggest you follow his advice.

Well, actually, I wrote really good software for mapping and variant-calling of Solid reads. But then I deleted them because, well, the platform was obsolete. And unfortunately, UTSW is very strict about preventing employees from providing knowledge to the rest of the world... their legal department told me very bluntly that all of the taxpayer- and grant-funded development that I did was their property, which they choose to keep secret.

So! The Solid platform is terrible. I wrote software that actually makes it useful. But I'm not allowed to share it with you.

Still, no matter how good the software is, Illumina is vastly better than Solid, so it's best to discard the Solid data and use an alternative platform.

gringer · 05-19-2017, 01:17 PM

I completely agree with pmiguel:

https://groups.google.com/d/msg/trin...c/xS3loMsNcpYJ

http://seqanswers.com/forums/showpos...56&postcount=4

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Strange Illumina fastq header	chariko	Bioinformatics	2	07-27-2016 04:59 AM
how to convert general fastq to fastq int format?	feng	Bioinformatics	21	07-04-2014 12:40 AM
For MAQ: Is there a Tool to convert sanger-format fastq file to illumina-fotmat fastq	byb121	Bioinformatics	6	12-20-2013 02:26 AM
strange Illumina txt format	m_elena_bioinfo	Bioinformatics	11	10-03-2013 09:15 AM
Question about using sra_toolkit to transform the SRA format into FASTQ format	areyousad	Bioinformatics	0	05-16-2010 11:56 PM

05-19-2017, 11:31 AM	#5
Brian Bushnell Super Moderator Location: Walnut Creek, CA Join Date: Jan 2014 Posts: 2,707	Haha But, yes, pmiguel is completely correct and I suggest you follow his advice. Well, actually, I wrote really good software for mapping and variant-calling of Solid reads. But then I deleted them because, well, the platform was obsolete. And unfortunately, UTSW is very strict about preventing employees from providing knowledge to the rest of the world... their legal department told me very bluntly that all of the taxpayer- and grant-funded development that I did was their property, which they choose to keep secret. So! The Solid platform is terrible. I wrote software that actually makes it useful. But I'm not allowed to share it with you. Still, no matter how good the software is, Illumina is vastly better than Solid, so it's best to discard the Solid data and use an alternative platform. Last edited by Brian Bushnell; 05-19-2017 at 11:55 AM.

05-19-2017, 10:20 AM	#2
mastal Senior Member Location: uk Join Date: Mar 2009 Posts: 667	It's data from the SOLiD platform. They use colorspace rather than basespace, so you get a sequence of 0,1,2,3 rather than A,T,G,C. The . indicate missing base calls.

05-19-2017, 11:13 AM	#4
pmiguel Senior Member Location: Purdue University, West Lafayette, Indiana Join Date: Aug 2008 Posts: 2,317	If you want to map those to a reference: (1) Don't. Delete the file and pretend like it never existed. (2) No, really, I mean it. Delete the file. (3) Okay, the reason there are so many "." in the sequence is that the SOLiD gave you reads from every bead it found, no matter how bad the data was. Also the .fastq's started at the very top of the slide, near the edge, where the data is worst. (4) But you don't need to know this, right? Please tell me you took my advice and deleted the file? (5) I mean, you can't just take the numbers and convert them to sequence. Any string of numbers could encode any of 4 sequences. It is the base at the beginning of each sequence that tells you the right conversion path. But if any "base" in the read was incorrect, it corrupts the conversion path. So if you did map these, you would want to use an old-time mapper in "colorspace" mode to do the mapping. BWA and Bowtie probably can still do this type of mapping. (6) But you don't need to know that because you deleted the file, right? If not, well, I did warn you... -- Phillip

05-19-2017, 01:17 PM	#6
gringer David Eccles (gringer) Location: Wellington, New Zealand Join Date: May 2011 Posts: 838	I completely agree with pmiguel: https://groups.google.com/d/msg/trin...c/xS3loMsNcpYJ http://seqanswers.com/forums/showpos...56&postcount=4