Dropbox has released an additional free 2GB space named as a "work Dropbox account" connected to your original account
To start it, you could quickly register from their webpage, following their tour.
After registering, you can sync it by doing this:
1. Click on the Dropbox icon -> click on the gear icon
2. Preferences... ->Account tab -> Buisiness tab
3. Link and log in
Wednesday, April 26, 2017
Monday, April 24, 2017
Reading a Json String on Linux/Unix Shell? Here's how I use jq
This prints out all the fields in your json string:
jq keys YourJsonString
This prints out the the field called "accession" of all your nodes:
jq '.files[].accession' YourJsonString
Usually, assuming the outputs follow a constant order, you can juxtapose the fields you need together to make a table by typing this:
paste -s <(jq '.files[].href' test) <(jq '.files[].accession' YourJsonString)New type of phishing email targeting Caltech students
From: Library Services - California Institute of Technology <libraryservices@hacettepe.
Date: Mon, Apr 24, 2017 at 9:43 AM
Subject: Library Services
To: brian.he@caltech.edu
Date: Mon, Apr 24, 2017 at 9:43 AM
Subject: Library Services
To: brian.he@caltech.edu
Dear Student,
https://clsproxy.library.calte
If you are not able to login, please contact Sarah Miller at shmiller@emich.edu for immediate assistance.
Sincerely,
Sarah MillerCalifornia Institute of TechnologyCaltech Library1200 E. California Blvd.Mail Code 10-90Pasadena, CA 91125T: 626-395-3405
https://clsproxy.library.calte
If you are not able to login, please contact Sarah Miller at shmiller@emich.edu for immediate assistance.
Sincerely,
Sarah MillerCalifornia Institute of TechnologyCaltech Library1200 E. California Blvd.Mail Code 10-90Pasadena, CA 91125T: 626-395-3405
Saturday, April 22, 2017
How to copy the hyperlink in a cell in Excel spreadsheet without opening it, on Mac
move your cursor to select that cell using your array keys
command + k
then command + c
command + k
then command + c
Thursday, April 20, 2017
Tuesday, April 18, 2017
Cufflink outputs FAILS
http://seqanswers.com/forums/showthread.php?t=14864
Cole Trapnell
A bit more on what FAIL means, and how it can happen. We use FAIL for genes that actually throw a numerical exception during isoform abundance calculation. In Cufflinks and Cuffdiff, there's a couple of calculations that require us to build matrices with either a row per transcript and a column per read (more or less) or a square matrix with a row and column for each transcript. Some of these matrices need to be invertible or positive definite or have other properties in order for the next steps in the algorithm to succeed. However, sometimes (due to things like round-off error) they aren't. Other times, missing data causes trouble. Oddly enough, this is actually more likely to happen the more reads you get overall, because you can see that isoforms are present, but you don't actually have enough data to calculate those abundances. This is the effect you were observing above. So since we can't be sure about the values (and in fact, were we to go ahead and do the calculation anyways, they could be *wildly* off in theory, or even negative), we set them to zero and move on.
In order to make differential expression estimates more conservative, version 1.1.0 really ramped up the checks that are done before these steps so we don't end up reporting false positives that are due to numerical exceptions. However, users (like yourselves) have been pretty frustrated by those changes, so I've spent the last few weeks going back and streamlining the overall algorithm to actually eliminate pieces that require the matrices to have some of those properties. The main offender was our "importance sampling" procedure, which tries to give us a sense (for each gene) for the accuracy for the maximum likelihood estimate of isoform abundances. This procedure was originally meant to improve the robustness of the estimate when one or more isoforms were close to zero, but in practice, we found that it actually hurts as often as it helps. Moreover, this procedure would often FAIL genes, so I removed it altogether. I've compensated on the differential expression side with some other statistical improvements and fixes, and the result is globally more accurate differential analysis (both in terms of fewer FAILs and fewer false positives than 1.1.0).
The upcoming version 1.2.0 should drastically reduce the number of FAIL genes, though there will still be some. If we can't calculate an MLE to begin with, or if for some reason the confidence interval calculation fails, a gene will be marked as FAIL.
Hope this sheds light on things.
Cole Trapnell
A bit more on what FAIL means, and how it can happen. We use FAIL for genes that actually throw a numerical exception during isoform abundance calculation. In Cufflinks and Cuffdiff, there's a couple of calculations that require us to build matrices with either a row per transcript and a column per read (more or less) or a square matrix with a row and column for each transcript. Some of these matrices need to be invertible or positive definite or have other properties in order for the next steps in the algorithm to succeed. However, sometimes (due to things like round-off error) they aren't. Other times, missing data causes trouble. Oddly enough, this is actually more likely to happen the more reads you get overall, because you can see that isoforms are present, but you don't actually have enough data to calculate those abundances. This is the effect you were observing above. So since we can't be sure about the values (and in fact, were we to go ahead and do the calculation anyways, they could be *wildly* off in theory, or even negative), we set them to zero and move on.
In order to make differential expression estimates more conservative, version 1.1.0 really ramped up the checks that are done before these steps so we don't end up reporting false positives that are due to numerical exceptions. However, users (like yourselves) have been pretty frustrated by those changes, so I've spent the last few weeks going back and streamlining the overall algorithm to actually eliminate pieces that require the matrices to have some of those properties. The main offender was our "importance sampling" procedure, which tries to give us a sense (for each gene) for the accuracy for the maximum likelihood estimate of isoform abundances. This procedure was originally meant to improve the robustness of the estimate when one or more isoforms were close to zero, but in practice, we found that it actually hurts as often as it helps. Moreover, this procedure would often FAIL genes, so I removed it altogether. I've compensated on the differential expression side with some other statistical improvements and fixes, and the result is globally more accurate differential analysis (both in terms of fewer FAILs and fewer false positives than 1.1.0).
The upcoming version 1.2.0 should drastically reduce the number of FAIL genes, though there will still be some. If we can't calculate an MLE to begin with, or if for some reason the confidence interval calculation fails, a gene will be marked as FAIL.
Hope this sheds light on things.
Load a bam file onto UCSC genome browser
track type=bam name=file1 bigDataUrl=http://link/to/your/bam/file
just be sure you have indexed the bam file.
from http://seqanswers.com/forums/showthread.php?t=11785
just be sure you have indexed the bam file.
from http://seqanswers.com/forums/showthread.php?t=11785
Subscribe to:
Posts (Atom)