ram processor storage and time info for characterizing networks 03-04-2015d1621

2015-03-06

-Is it possible to analyze all of the metrics (except for search information entropy; so n	m	avg_shortest_path	diameter	clustering_coefficient	power_law_exponent) constructed from a gene co-expression network of 64,000 genes in a reasonable amount of time?
--
{
estimates on Samsung Ativ Book 6 Laptop
{
features	n	m	time (s)	metrics calculated	note
250	247	4799	5	"n	m	avg_shortest_path	diameter	clustering_coefficient	power_law_exponent"	-
300	300	6782	9	"n	m	avg_shortest_path	diameter	clustering_coefficient	power_law_exponent"	-
1000	1001	72934	>7*60	"n	m	avg_shortest_path	diameter	clustering_coefficient	power_law_exponent"	It's the clustering coefficient that is taking so long to calculate; maybe this can be approximated as well
1000	1001	72934	82.112	"n	m	avg_shortest_path	diameter	power_law_exponent"
10000	10000	-	>18*60	-	With 10,000 features, the program starts outputting all of the data into multiple text files. . .
10000	10000	8203987	>12*60*60	-	I tried to just calculate search information entropy with just a 1,000 node pairs, but I don't think the all pairs shortest paths was calculated in a reasonable time
}
note that I have seen that the eixe cnio computer is faster than my samsung ativ book 6 computer, but it still takes quite a while to try to evaluate the network from 72,934 features

estimates on eixe computer
{
features	n	m	time (s)	metrics calculated	note
1000	1001	72934	1418.001	"n	m	avg_shortest_path	diameter	clustering_coefficient	power_law_exponent"
}
}

azim58wiki: