PS8.1: Write an outline that describes the strategy employed by the subroutine print-context within find-context.plto locate the appropriate orfs for each hit in motif.hits. What assumption(s) does the routine make? (Since the program works, these assumptions are evidently correct)

PS8.2: The program find-context.pl assumes that memory is not a limiting resource and proceeds to read into memory all the information from the data file it will ever need. This is often not a bad assumption, but as we want to consider more and more information it can become bad. Suppose that you’re working on a PC with 64 Mbyte memory (call me old-fashioned, but that’s my PC), and, for safety reasons, you decide you don’t want to use more than 32 Mbytes of it on running this program.
 

PS8.2a. Calculate how much memory you’re using in storing all the information you read from 7120DB.dat. It's not always easy to tell how much memory Perl will use for its data.  Try something like this: 20 bytes for each number, 25 bytes for each string, plus 1 byte for each character in the string, and 72 bytes for each array (where each row [ ... ] counts as an array). You may find that you’re still lacking one key piece of information needed to calculate this number. If so, then write a quick program (actually, alter find-context.pl) so that it gives you this piece of information.

PS8.2b. Suppose that you want to run the program using a similar database file that covers all human genes. Now how much memory do you need?

PS8.2c. Suppose that want the program to give you not only the coordinates of the relevant genes but also the 3000 bp adjacent to each of the identified orfs. This would be important if you wanted to investigate the sequences of complex regulatory sequences governing the expression of the genes you identify. NOW how much memory do you need?

• Read 7120DB.dat sequentially, storing only the information you will need to find the orfs you need, reading it into an array (What do you read? How do you store it?).
• Read motif.hits sequentially, considering one hit at a time (What do you read? How do you store it?)
• For each hit, find the number of the record you want. To do this, go through the array you created until you find the orf you want, then consider the subscript of the array to be the number of the desired record. (describe more precisely how you find the desired orf).
• Calculate where in 7120DB.dat (how many bytes into the file) is the desired record (describe this calculation).
• Read the number of bytes you want starting with the byte number you calculated above. (You don't know any Perl statement that can do this, but assume one exists)

• Read motif.hits sequentially, considering one hit at a time.
• For each hit, search through 7120DB.dat until you find the desired record.

PS8.4a. Presuming that you search 7120DB.dat sequentially, what is the average number of records you’ll have to go through to find some random coordinate read from motif.hits?

PS8.4b. Suppose that searching through this number of records leads to unacceptably long waits (it does on my machine) (with my level of patience). Devise a search strategy that cuts the search time down enormously. [Hint: consider how you would find a name in the phone book – not by scanning from page 1!]

PS8.4c. There are (at least!) two such strategies. One benefits from assuming that motif.hits is sorted (it's not, but clearly we could sort it and print it out again). Another strategy works whether or not motif.hits is sorted. Which was yours? What would the other one look like?

Can't yet print context without right-hand ORF

PS8.5a. Why?
PS8.5b. Modify find-context.pl so that it works properly with test-motif.hits.

<--*all0606*PetC: cytochrome b6/*104*704137*704212*467*-->*alr0607*NirA: nitrite reduct*25.5

PS8.6a. Modify find-motif.pl so that it gives the desired output.

PS8.6b. Examine the output in Excel. What motifs look interesting? By “interesting” I mean that they are placed upstream from at least one gene of known function. To facilitate your search, sort the data so that hits between  two upstream regions are at the top of the list, then those with one upstream region, then those with no upstream region (or the NtcA site is within a gene).

PS8.7: Consider the unpack statements (related to input) and printf statements (related to output) you have run across. Never mind graphic input and output, which add another order of magnitude of complexity. How much easier would programming be if you could let someone else worry about input and output?