Hey Rohit, what’s happening NOW?
So a website with a link that says “now” goes to a page that tells you what this person is focused on at this point in their life. For short, we call it a “now page”.
Note: Periodic updates from this NOW page can also be retrieved by finger rohit@rohitfarmer.com
Location
North Bethesda, Maryland, USA
Non-tech/personal
On going thoughts & status
These statuses are toots from my Mastodon profile at https://mindly.social/@ro
Periodic updates
May 1, 2024
April was filled with fun and adventure whenever it was possible. Elsa was tasked with taking care of her class pet, Hoot, for a weekend. Hoot is a beloved owl (stuffed toy) who goes to one of the kids’ homes every weekend and does activities with them. The activities are logged in a notebook and sent back to the class teacher.
On that weekend with Hoot, we went Geocaching, attended a birthday party, met a ninja at the party, and learned some ninja moves. We also baked some cookies with Hoot. Printed pictures with Hoot were placed in the log book, and Elsa wrote the details of the activities.
Then came the solar eclipse. Since we were not on the path of totality, we had no plans to view it, but at the last moment, we decided to do so and somehow got some eclipse glasses. Trust me, it was hard to find them. Sibyl had to go to work in person, so she watched it from there. I worked from home and saw it from my balcony. We decided not to send kids to school as the timing coincided with their departures and outdoor breaks, and we were not sure how school or daycare would manage kids not looking into the sun without proper care. Kids and I saw some programming on TV about how solar eclipses happen, and then I took Elsa out on the balcony to see the partial eclipse. I witnessed a total solar eclipse back in India when I was a teenager, so I remember it vividly; therefore, it was not as moving for me as for other folks who saw such a heavenly phenomenon like this for the first time. One of my colleagues drove 9 hrs from Maryland to further north to put himself on the path of totality. He was amazed.
Elsa has finally launched her YouTube channel Elsatastic. It took her lazy parents one year to finally set it up for her. She already has a couple of videos and loves watching herself on the big-screen TV. Please subscribe to her channel. It will make our little girl very happy.
We also resumed our fishing adventure on an unexpectedly warm sunny day at Pine Lake in Wheaton Regional Park. For the first time, I caught a considerably large Bluegill with a lure. We decided to bring it home, and I wanted to cook it, but my kids did not allow me to. So once it died (sadly), I trashed it and told the kids that it jumped into the toilet and swam its way back to the lake.
We ended the month with a one-night camping trip at Cedarville State Forest. It was our first camping trip of the season and Elsa and I were joined by our neighbors, Vaughn, his son, Harish, and his family. Sibyl and Eva decided to stay back. We started on Saturday at 7 a.m., reached the campsite at 9 a.m., and checked out on Sunday at noon. After spending last year curating my camping gear, this time, I had everything I needed, and therefore, I enjoyed the camping itself instead of stressing about the stuff that I did not have or did not know how to use it. We cooked chicken fajita wraps, chili mac and cheese, lots of coffee, corn on the cobs, sweet potato, and, of course, s’mores.
Tech/professional
On going thoughts & status
These statuses are toots from my Mastodon profile at https://fosstodon.org/@swatantra
Periodic updates
May 1, 2024
In April, I mostly worked on metagenomics and PhIP-Seq data from two projects. I utilized CZID for metagenomics, which carried out host genome depletion and provided water contamination information in the form of z-scores per taxons that it had identified in the samples. Although CZID’s website is excellent, it was not clear that the water contamination depletion is not reflected in read counts but is presented in the form of z-scores, as mentioned above. We realized it when we saw the enrichment of taxons that did not make biological sense in the downstream analysis. Z-scores can only be exported sample-by-sample along with other information, such as taxonomic IDs. I had to write a function to combine these multiple sample-by-sample files into a combined file while filtering the taxons that did not meet our z-score cutoff. Filtering taxons using the z-scores has eliminated some of those problematic taxons that were seen as highly enriched.
For the PhIP seq analysis, we ran into an issue with another dataset that was generated using a smaller library/panel. Many of the tiles did not receive a p-value in the regression analysis, and it was not apparent why. My theory is that the phipstat code is performing several divisions, and in one of those divisions, the values are going smaller than what R can handle natively; hence, the values are rounded to zero, which is then passed to a log function, which returns an Inf value thus breaking the pipeline. I crudely fixed it to stop the optimizer in regression as soon as it hits an Inf value. This workaround does not break the pipeline, but for many tiles, the optimization terminates before convergence. A solution could be to allow R to handle long doubles using a package like Rmpfr. Another solution is to use larger tile libraries in our experiments. What is the sweet spot? I do not know.