parsing the data

[Kirthi Pulakanti]

Hello ,

I have a data in format :

CHR	TSS-25bp	TSS+25bp	count	tss	Ensemble transcript	refgene	strand
chr8	68141773	68141823	1	68141798	ENSMUST00000152320	1-Mar	+
chr8	68141882	68141932	3	68141907	ENSMUST00000110258	1-Mar	+
chr8	68141898	68141948	3	68141923	ENSMUST00000110256	1-Mar	+
chr8	68141910	68141960	3	68141935	ENSMUST00000155804	1-Mar	+
chr8	68141959	68142009	2	68141984	ENSMUST00000110255	1-Mar	+
chr8	68910167	68910217	2	68910192	ENSMUST00000039540	1-Mar	+
chr8	68910174	68910224	2	68910199	ENSMUST00000110253	1-Mar	+
chr17	33822631	33822681	2	33822656	ENSMUST00000066121	2-Mar	-
chr17	33828434	33828484	2	33828459	ENSMUST00000172767	2-Mar	-
chr17	33828758	33828808	1	33828783	ENSMUST00000173454	2-Mar	-
chr17	33840058	33840108	1	33840083	ENSMUST00000173392	2-Mar	-
chr18	56963297	56963347	1	56963322	ENSMUST00000153044	3-Mar	-
chr19	37282007	37282057	4	37282032	ENSMUST00000024078	5-Mar	+
chr19	37282032	37282082	5	37282057	ENSMUST00000112391	5-Mar	+
chr19	37282040	37282090	4	37282065	ENSMUST00000148105	5-Mar	+
chr15	31385628	31385678	2	31385653	ENSMUST00000090227	6-Mar	-
chr15	31387011	31387061	1	31387036	ENSMUST00000043826	6-Mar	-

For every ref genes ,I would like to get a unique count (column 7) on the basis of strand(column8) and tss( column5) information. For example,

There are 7 rows for 1-March with count 1,2,3 .If a gene has same number of count, I would like to retrieve that gene with its highest number for tss, likewise if its on negative strand ,keep the one with lowest tss number (starting from 3'). For gene 1- March and 2-March, I would expect;

chr8

68141773

68141823

1

68141798

ENSMUST00000152320

1-Mar

+

chr8

68141910

68141960

3

68141935

ENSMUST00000155804

1-Mar

+

chr8

68910174

68910224

2

68910199

ENSMUST00000110253

1-Mar

+

chr17

33822631

33822681

2

33822656

ENSMUST00000066121

2-Mar

-

chr17

33828758

33828808

1

33828783

ENSMUST00000173454

2-Mar

-

Sorry for the typo,it should be gene March1 and March2 . 

CAN YOU PLEASE HELP ME GIVE IDEAS OF HOW DO I START WRITING A SCRIPT IN UNIX . ?

Thanks for your time. 

[Keith Bradnam]

Some more information might help.

1. How many rows in your input file?
   
2. Does every row contain a unique ID in the Ensembl transcript column?
3. Is the input file sorted by ref gene, or could you expect to see '1-Mar' further down the file?

Have you tried any approaches yourself yet?

Regards,

Keith