| 1) |
FTP from Cedar or HTTP from
http://www.som.soton.ac.uk/research/geneticsdiv/epidemiology/LDMAP/
|
| 2) |
From ver. 29Apr06 onward,
LDMAP_cluster performs
automatic online update. This ensures the latest version of the program is
delivered to all users. (29Apr06).
If you are still using the older version of program, you could check the
version date on the program header
with the version date on the website here.
|
| 3) |
LDMAP_cluster program is able to handle
multi-dataset submission from ver. 10Apr06 onward. It is recommend to
submit 2 datasets (~60k SNPs each) at the same time to avoid exceeding
the 5Gb storage space.
However if already have 2 datasets
running, and have checked you have more than 3Gb left ("checkseg" and "du -s ../<username>", by all means
submit another one. Each dataset consumes ~ 2 - 2.5Gb.. depending on the
# of SNPs the dataset has.
|
| 4) |
You are given 5Gb of storage space on Iridis 2, use
"du -s ../<userID>" to
check available space, click
here.
|
| 5) |
The time for constructing a LDU map from SNP dataset
is influenced by many factors such as the size of the dataset, the
availability of computing resource on Iridis 2. Currently (27Apr06),
the maximum limit of this program is ~70,000 SNPs. If a dataset contains
>70k SNPs, it is split into smaller datasets with an overlap of ~200
SNPs. Under
favourable
conditions, the LDU map could be constructed in 3 - 5 hrs from a single
dataset of 68k SNPs (34 segments running in parallel).
Typically under
the normal usage load on Iridis 2, you
could expect to have the results in 24 hrs.
|
| 6) |
The segment / job is terminated (killed) from the
computer cluster once the
requested time (10 hours) is up.
This is very rare, only 7 segments has exceeded the 10 hrs out of 4195
segments. Please contact me if you encountered such instance. Amendment
on the submission script is necessary to request a longer computing
time.
|
| 7) |
No further submission of dataset is possible, and
the running jobs will not be able to create or write any new result to
the existing files. Use "checkSeg"
to free more space.
|
| 8) |
Use "checkSeg", "qstat
-a" or "showq", click
here. A 2000 SNPs segment (excl. extended / contig region) takes ~2 - 5 hrs, however some segments do exceed the 10
hrs computing time. Please click here
for more details.
|
| 9) |
Under rare
circumstances, there might be job(s) or segment(s)
shown as "Idle" among the completed segments on the list from "checkSeg".
The "idle" segment(s) will not be found on either "qstat -a" or "showq". This is accounted for by the
segment(s) not being submitted / executed by the PBS. So far such
incidents have been reported 4 times out of the
4195 segments / jobs.
The LDU map cannot be
assembled if anyone of the segments is missing. There are two ways to
resolve this problem. One way is to delete all the existing results and
re-run the LDMAP_cluster with the same dataset. The other way is to save /
transfer the results of the completed segments ("retrieveResult") and manually
re-submit the "idle" segment(s) to Iridis.
It is highly recommended to re-submit the idle one(s) only rather than
re-computing all the segments again. The simplest way is to look for the
submission script associated to the "idle" segment (e.g. idle segment =
segment 5 in dataset JPT6_3,
then the submission script you need is "jpt6_3.dat_seg_5_6" located at the home directory).
Once located the submission
script, re-submit the job via "qsub"
manually from the home
directory.
Although an additional segment (i.e. 6) is re-submitted along the "idle"
segment 5, it is better than amending the submission script as below:
If you really want submit the job with a single segment, please read
on... Since LDMAP_cluster embeds two jobs (segments) for each submission,
it is necessary to amend the
script. Below is an example of the script for a single job for
submitting segment 5.
#!/bin/sh
#PBS -l walltime=10:00:00
#PBS -l nodes=1
#PBS -N jpt6_3.dat_seg_5
cd jpt6_3.dat_seg_5
ldmapper1+ jpt6_3.dat job
int_5 ter_5 pro_5 7900 10100
|
|
| 10) |
The segment(s) / job(s) has/have exceeded the 10
hrs computing time and has/have been removed from the cluster. It is
quite rare for a segment to take longer than 10 hrs to compute.
Examples have been reported in 7 out of 4195 segments. Please contact
me if you encounter such an instance. From 19May06 version onward, the
window size has been optimized to 75 SNPs, there should not be any
segment(s) / job(s) to exceed 10 hours computing time.
The segments that have exceeded 10 hrs computing time are:
dataset chb6_2 segment 20;
dataset jpt6_2 segment 20;
dataset jpt6_3 segment 13;
dataset ceu8_2 segment 17 and 18;
dataset jpt8_2 segment 16;
dataset ceu11_1 segment 23
|
| 11) |
This is more likely to be associated with the
dropped SNP, please click here
for further
discussion. Currently this is fixed manually. Please contact me if you
encounter such an instance.
|
12)
|
Yes, all
researchers or research students of Southampton University would have
access to LDMAP_cluster. Please click here for
more details.
|
13)
|
Currently, it is not possible for
external members of Southampton University to access LDMAP_cluster
on
Iridis cluster at the University. A web portal is planned to extend the
accessibility of LD analysis as a service. Alternatively, you could run
the program on your Beowulf cluster.
|
| 14) |
Yes, please click
here
for more information.
|
15)
|
No. The chromosome-wide LDU map is
constructed employing the
segmental approach, where the chromosome is
split into segments of 2000 SNPs, and an epsilon is estimated in every segments (pro
files).
Please note that although it is a segment of 2000
SNPs, the actual number of SNPs in each segment is greater. The extra
SNPs are accounted for the
extended regions at the split end(s) of the
segment: 2100 SNPs
at the first segment; 2200 SNPs at the middle ones and a variable number
of SNPs (1000 - 2999) at the last segment, click
here for further information.
So, the epsilon in each pro file represents a
region of >2000 SNPs.
The
number of SNPs for the end segment can be inspected in the file, "<datasetfile>_contigInfo.txt",
under the result directory.
(e.g.
\result_ceu1_3.dat_1145206991\ceu1_3.dat_contigInfo.txt). Click
here to view an example of the
file.
|
