rsem - find number of reads in original data

[kmkocot]

Hi all,

I ran rsem on a bunch of Illumina transcriptomes to get FPKM values. Because of the large file size, I deleted the .fastq files containing the raw reads from my local machine. Now I realize that I never counted the number of reads in each input file and need to put that information in a poster I'm making without having to re-download the raw data (it's on a server in the US and I'm in Australia). I was wondering if there is an easy way to glean the original number of reads from from the rsem output files? I figured the .cnt file would but I compared the values in this file to the raw read count for one dataset that I re-downloaded and I can't figure out a correlation.

The .cnt file for one dataset:

23749538 2281663 0 26031201
987449 1294214 1331390
6796612 3
0	23749538
1	950273
2	505959
3	212225
4	250642
5	92501
6	87457
7	20239
8	48846
9	13012
10	32131
11	23953
12	8958
13	656
14	5164
15	3246
16	4698
17	337
18	2680
19	101
20	2745
21	1387
22	2533
23	280
24	2154
25	452
26	293
27	264
28	350
29	233
30	1883
31	87
32	355
33	763
34	64
35	285
36	61
37	29
38	255
39	22
40	860
41	50
42	76
43	5
44	75
45	26
46	1
47	6
48	183
49	11
50	116
51	3
52	50
53	3
54	32
55	7
56	13
57	73
58	3
59	6
60	1511
61	6
62	4
63	13
64	42
65	136
66	92
67	28
68	11
69	122
70	3
71	2
72	1
74	2
75	1
76	3
77	1
78	1
80	2
81	1
82	1
83	1
87	1
89	1
90	3
92	1
93	1
94	2
95	3
96	4
97	4
98	13
99	2
100	77
101	392
102	3
Inf	0

There were 26031201 pairs of reads that went into this rsem analysis.

Thanks!
Kevin

[TiborNagy]

If you keep the BAM files you can count the raw read numbers.

[kmkocot]

Will this count all of the reads in the input .fastq files or just the reads that mapped? I think the latter is the case, right? Is there a quick script / tool you can recommend to do this in batch?

[dpryan]

That'll depend on the aligner and the settings. You can do a quick

samtools view -cf 0x4 blah.bam

to see if your aligner kept unaliged reads in the BAM file (if the output is more than 0 then this is the case).