Contact
I'm happy to discuss research collaborations, project supervision, consulting, and reproducible computational workflows. The fastest way to reach me is by email.
Preferred contact
dipankatanusarmah@gmail.comFor project-related messages, a short dataset summary + goals + timeline helps me respond faster.
Submit a pitfall
Have a bioinformatics "gotcha" you've seen in the wild? Send it and I can add it to the Debug Your Science collection.
Tip: Include a short example (code snippet or plot) and what the correct fix looks like.
Debug Your Science
Common pitfalls that turn good data into bad conclusions. Click any card to reveal the fix.
The Excel Gene Name Trap
Mistake: Importing gene lists into Excel. Gene symbols like SEPT6 or MARCH1 get auto-converted into dates (e.g., “6-Sep”, “1-Mar”).
Result: Silent corruption of gene names → wrong hits, wrong enrichment, irreproducible results.
Ignoring Adapter Sequences
Mistake: Skipping adapter/quality trimming before alignment or quantification.
Result: Lower alignment rates, biased counts, and artifacts that look like real variants or splicing.
The Multiple Testing Oversight
Mistake: Running thousands of hypothesis tests and treating p < 0.05 as significant without correction.
Result: Hundreds to thousands of false positives by chance alone.
Batch Effects Disguised as Biology
Mistake: Case/control samples processed on different days, lanes, kits, or operators without randomization.
Result: You model the sequencing run, not the disease signal.
The Hairball Visualization
Mistake: Plotting 5,000 nodes and 50,000 edges and calling it insight.
Result: A beautiful mess: no interpretation, no prioritization, no story.
Correlation ≠ Interaction
Mistake: Assuming co-expression implies physical interaction or regulation.
Result: False mechanistic claims; wrong targets prioritized.
Hard-coded File Paths
Mistake: Code that only works on your machine: C:/Users/John/Desktop/data.csv
Result: Nobody can reproduce it — including future you.
Reference Genome Mix-up
Mistake: Aligning to hg19 but interpreting coordinates from GRCh38 (or vice versa).
Result: Mis-mapped loci, wrong annotations, invalid variant interpretation.